indicate the location of the instruction

Abstract

The present invention relates to indicating locations of instructions, and specifically discloses apparatuses, systems, and techniques for causing one or more storage locations of one or more scheduled instructions to be indicated to one or more users. In at least one embodiment, one or more storage locations of one or more scheduled instructions are caused to be indicated to one or more users based on, for example, an application programming interface (API).

Description

[0001] Cross-reference to related applications For all purposes, this application incorporates by reference the entire disclosure of co-pending U.S. Patent Application No. 19 / 181,203, entitled “Preventing Indication of Locations of Instructions,” filed concurrently with this document, and U.S. Patent Application No. 19 / 015,274, entitled “Preventing Instructions from Being Performed,” filed January 9, 2025. Technical Field

[0002] At least one embodiment relates to processing resources used to instruct instructions to be scheduled for execution. Background Technology

[0003] In computing systems with multiple processing units, managing the submission of tasks to be executed by those units is crucial for maintaining high performance and control. Managing these tasks can be complex and resource-intensive, especially when tasks have already been queued by one processing unit for execution by another, where task information may not be readily available. The amount of memory, time, or computational resources used to manage the submission of tasks to be executed by processing units can be improved. Attached Figure Description

[0004] Figure 1 The illustration shows an example computing system for monitoring submitted tasks according to at least one embodiment; Figure 2 The illustration shows an example of a portion of a process for monitoring a submitted task according to at least one embodiment; Figure 3 The illustration shows an example of a portion of a process for monitoring a submitted task according to at least one embodiment; Figure 4 The illustration shows an example of a portion of a process for monitoring a submitted task according to at least one embodiment; Figure 5 An example of an API for initiating a monitoring task, according to at least one embodiment, is illustrated; Figure 6 An example of an API for stopping a monitoring task, according to at least one embodiment, is illustrated; Figure 7 The illustration shows an example of a process for indicating the location of a scheduled instruction according to at least one embodiment; Figure 8 An example data center system according to at least one embodiment is shown; Figure 9 A system-on-a-chip (SOC) according to at least one embodiment is shown. Figure 10A A parallel processor according to at least one embodiment is shown; Figure 10B A processing cluster according to at least one embodiment is shown; Figure 10C A graphics multiprocessor according to at least one embodiment is shown; Figure 11 An accelerator processor according to at least one embodiment is shown; Figure 12A A central processing unit according to at least one embodiment is shown; Figure 12B The illustration shows an embodiment according to at least one of the embodiments. Figure 12A The core of the central processing unit; Figure 13 Another accelerator processor according to at least one embodiment is shown; Figure 14 A neuromorphic processor according to at least one embodiment is shown; Figure 15 A supercomputer according to at least one embodiment is shown; Figure 16 Another accelerator processor according to at least one embodiment is shown; Figure 17 Another processor according to at least one embodiment is shown; Figure 18 Another accelerator processor according to at least one embodiment is shown; Figure 19 A tensor processing unit according to at least one embodiment is shown; Figure 20 A RISC-V compatible processor according to at least one embodiment is shown; Figure 21A and Figure 21B A language processing unit according to at least one embodiment is shown; Figure 22 A software stack of a programming platform according to at least one embodiment is shown; Figure 23 Software supported by a programming platform according to at least one embodiment is shown; Figure 24 A method for using at least one embodiment is shown. Figure 23 Compiled code executed on the programming platform; Figure 25 An example of an autonomous vehicle and its system architecture according to at least one embodiment is shown; Figure 26A The inference and / or training logic according to at least one embodiment is illustrated; Figure 26B The inference and / or training logic according to at least one embodiment is shown; and Figure 26C Training and deployment of a neural network according to at least one embodiment are illustrated. Detailed Implementation

[0005] In one example, the technique described herein improves program execution in a computing system by providing the user with indications of the location and other information of the instructions used to execute a task scheduled by the central processing unit (CPU) for execution on a graphics processing unit (GPU). This addresses a challenge that users may face when scheduling tasks using the CPU to execute on the GPU, where the user may expect to modify or prevent the task from executing after it has been scheduled for execution on the GPU. By providing the user with information about the instructions used to execute the scheduled task (such as their location in memory), the computing system can improve the user's control over the execution of those tasks, as the user can use this information to modify or prevent the scheduled task from executing.

[0006] The techniques described herein can relate to a computing system that includes mechanisms for monitoring tasks queued for execution by a GPU but not yet picked up or otherwise processed by the GPU's scheduler unit. Users can specify tasks in their programs, whereby users can insert markers at key points in the program to define task boundaries. The computing system can then store information about the task when it is scheduled for execution. In some embodiments, the computing system incorporates a push buffer to store information about the task, which can refer to a memory region shared between the CPU and GPU, where tasks are queued before execution. In at least one embodiment, a push buffer, as described herein, refers to any suitable memory region where tasks are queued before execution and can be referred to as a buffer, command buffer, circular buffer, indirect buffer, job queue, and / or variations thereof.

[0007] The computing system can track or otherwise monitor all tasks between user-inserted markers and store information such as the start and end addresses of the push buffer for the task's initiation method and all relevant semaphore values ​​that may be released upon task completion. In some examples, the computing system generates or otherwise utilizes a tracking object, which is a data structure that indicates or otherwise encodes information about the scheduled tasks and is accessible to the user. In one embodiment, the computing system implements or otherwise provides one or more application programming interfaces (APIs) that a user program can use to cause the computing system to begin storing information about any subsequently enqueued tasks and to stop storing information about any subsequently enqueued tasks. Specifically, the computing system can implement or otherwise provide APIs for indicating one or more storage locations of one or more scheduled instructions to one or more users, and APIs for preventing one or more storage locations of one or more scheduled instructions from being stored.

[0008] Numerous specific details have been set forth in the preceding and following description to provide a more thorough understanding of at least one embodiment. However, it will be apparent to those skilled in the art that the inventive concept can be practiced without one or more of these specific details.

[0009] Figure 1 An example computing system 100 for monitoring submitted tasks is illustrated according to at least one embodiment. In one embodiment, the computing system 100 includes one or more host processing units 102, a buffer 104, one or more device processing units 106, and / or a trace data structure 108. The computing system 100 can monitor tasks submitted by one or more host processing units 102 to one or more device processing units 106 via the buffer 104. In some examples, the computing system 100 uses the trace data structure 108 to store information about the scheduled instructions in the buffer 104. The computing system 100 can be combined with... Figure 2-7 The described computing system is consistent.

[0010] In one embodiment, computing system 100 is any suitable system and / or system (such as combined) Figure 8-2This is part of a set of systems described herein. Computing system 100 may implement or otherwise execute one or more driver computer programs. Computing system 100 may include circuitry and instructions for performing processes such as those described herein. In some embodiments, computing system 100 is associated with one or more programming models, such as the Unified Computing Device Architecture (CUDA) model, the Portable Heterogeneous Computing Interface (HIP) model, the oneAPI model, various hardware accelerator programming models, and / or variations thereof. Computing system 100 may be associated with any suitable machine learning and / or neural network framework, such as TensorFlow, Boost, Caffe, Microsoft Cognitive Toolkit / CNTK, MXNet, Chainer, Keras, Deeplearning4j, and / or variations thereof. Computing system 100 may include any suitable processing unit and / or combination of processing units, such as one or more central processing units (CPUs), graphics processing units (GPUs), and / or any suitable processing units.

[0011] In one embodiment, host processing unit 102 is any suitable one or more processing units, such as one or more CPUs, capable of implementing or otherwise executing one or more programs. Host processing unit 102 may include any suitable processing unit that can be referred to as a host in the context of one or more programming models. Host processing unit 102 can execute user programs, which can refer to any suitable software program, application, system, or module that can be executed using one or more processing units (e.g., one or more host processing units 102 and / or one or more device processing units 106), and may also be referred to as a user, user application, user software, user code, and / or variations thereof. User programs may be generated or otherwise obtained in combination with one or more users, systems, applications, and / or variations thereof. User programs may instruct a set of processes, which may be implemented as a set of instructions and may be executed or otherwise utilized by one or more host processing units 102 and one or more device processing units 106. In some embodiments, a user program executes one or more processes by causing at least one or more circuits (e.g., one or more processing units, systems and / or variations thereof, such as the processing units, systems and / or variations thereof described herein) to perform one or more processes.

[0012] User programs can instruct tasks to be performed by one or more device processing units 106. In one embodiment, one or more device processing units 106 are any suitable one or more processing units capable of implementing or otherwise executing one or more programs, such as one or more GPUs. One or more device processing units 106 can include any suitable processing unit that can be referred to as a device in the context of one or more programming models. In one embodiment, a task, such as the task described herein, refers to any suitable computation, process, and / or operation that can be scheduled to be performed by one or more device processing units 106, and may include one or more kernels, work items, methods, and / or commands implemented using any suitable data or format.

[0013] A user program may instruct a task to be executed by one or more device processing units 106. The task may be a kernel to be executed by one or more device processing units 106. As part of executing the user program, one or more host processing units 102 may cause the task to be queued into a buffer 104 for execution on one or more device processing units 106. In one embodiment, to queue a task into the buffer 104, one or more host processing units 102 cause instructions and / or other information associated with the task to be stored in the buffer 104. This instructions and / or other information may indicate any suitable information associated with the execution of the task, such as references to the task's code, grid and block dimensions, arguments (e.g., internal core arguments), resource requirements, memory references, synchronization information, and / or variations thereof. In the buffer 104, the instructions and / or other information may be referred to as scheduled instructions. In some examples, in order to enqueue a task into buffer 104, one or more host processing units 102 obtain or otherwise generate instructions required by one or more device processing units 106 to execute the task, and store these instructions in buffer 104 to schedule them for execution by one or more device processing units 106 or for other use, wherein the instructions in buffer 104 may be referred to as scheduled instructions.

[0014] In one embodiment, buffer 104 is a shared memory region between one or more host processing units 102 and one or more device processing units 106. Buffer 104 can be a memory region accessible to one or more host processing units 102 and one or more device processing units 106. In some embodiments, buffer 104 is also referred to as a push buffer, command buffer, circular buffer, indirect buffer, and / or variations thereof. Buffer 104 can be any suitable memory region for enqueuing tasks or otherwise instructing tasks for execution by one or more device processing units 106. Buffer 104 can be a memory region for temporarily storing scheduled instructions corresponding to tasks to be executed by one or more device processing units 106. In some examples, tasks can be enqueued into buffer 104 by storing at least the instructions corresponding to the task in buffer 104. Buffer 104 can be implemented as a first-in, first-out (FIFO) data structure. In at least one embodiment, one or more device processing units 106 include a scheduling unit that can obtain or otherwise process instructions and / or associated information stored in a buffer 104, such that the device processing unit 106 executes a task corresponding to the instructions and / or associated information. A scheduling unit or scheduler (such as the scheduling unit or scheduler described herein) can refer to any suitable set of hardware and / or software computing resources (such as an execution scheduler (ESCHED), a microengine scheduler (MES), and / or variations thereof) that can be used to schedule work to be executed on one or more processing units.

[0015] This allows computing system 100 to track or otherwise monitor tasks enqueued into buffer 104 as a result of a user program. The user program may include: invoking a start monitor API (e.g., [API name missing]) that causes computing system 100 to begin tracking enqueued tasks. Figure 5 The start monitor API 502) and the stop monitor API that can cause computing system 100 to stop tracking queued tasks (e.g., Figure 6The user program can cause the computing system 100 to store information about scheduled instructions in buffer 104 in a trace data structure 108, which can be any suitable data structure or object that can store or otherwise indicate information about queued tasks, scheduled instructions, and / or variations thereof. The computing system 100 can execute the start monitor API to cause the computing system 100 to begin storing or otherwise indicating information about scheduled instructions in buffer 104 (such as one or more memory addresses of scheduled instructions in buffer 104, one or more semaphore values ​​associated with scheduled instructions (e.g., to be released upon completion or execution of the scheduled instructions), one or more addresses of semaphores associated with scheduled instructions, and / or variations thereof), and can execute the stop monitor API to cause the computing system 100 to stop storing or otherwise indicating such information.

[0016] As an illustrative example, the user program includes invoking a start monitor API to cause the computing system 100 to begin storing information about scheduled instructions in buffer 104. Continuing this example, the user program instructs a first task to be executed by one or more device processing units 106, one or more host processing units 102 enqueuing the first task by storing a first set of scheduled instructions and / or associated information corresponding to the first task in buffer 104, wherein the computing system 100 also stores information about the first set of scheduled instructions and / or associated information in a trace data structure 108. Also continuing this example, the user program includes invoking a stop monitor API to cause the computing system 100 to stop storing information about scheduled instructions in buffer 104. Also continuing this example, the trace data structure 108 can be used to locate the first set of scheduled instructions and / or associated information in buffer 104, such as to modify or prevent the execution of the first task.

[0017] Figure 2 The illustration shows an example 200 of a process for monitoring submitted tasks according to at least one embodiment. In one embodiment, Figure 1 The computing system 100 monitors tasks submitted by one or more host processing units 102 to one or more device processing units 106 using buffer 104. User programs can instruct one or more kernels to be submitted by one or more host processing units 102 for execution by one or more device processing units 106.

[0018] The user program may include a start monitor API that enables computing system 100 to trace or otherwise monitor tasks enqueued into buffer 104. The user program may also cause the generation of a trace data structure 108. Computing system 100 may execute the start monitor API to cause computing system 100 to begin storing or otherwise indicating information about the tasks enqueued into buffer 104, such as one or more memory addresses of scheduled instructions in buffer 104, one or more semaphore values ​​associated with the scheduled instructions (e.g., to be released upon completion or execution of the scheduled instructions), one or more addresses of semaphores associated with the scheduled instructions, one or more channels associated with the scheduled instructions, and / or variations thereof. More information about the API can be found at [link to API documentation]. Figure 5-6 It was found in the description.

[0019] In one embodiment, a user program instructs kernel 202 to be executed by one or more device processing units 106. As part of execution by one or more host processing units 102, the user program can cause one or more host processing units 102 to submit kernel 202 to be executed by one or more device processing units 106, wherein kernel 202 may first be enqueued or otherwise stored in buffer 104. Kernel 202 may be enqueued, instructed, or otherwise stored in buffer 104 as a first instruction set and / or associated information, which can be implemented using any suitable data or format. As a result of submitting kernel 202 by one or more host processing units 102, computing system 100 can generate information about kernel 202 and store that information in trace data structure 108. The computing system 100 can monitor the process of queuing the kernel 202, such as the process of generating a first instruction set and / or information associated with the kernel 202 and / or storing it in the buffer 104, to generate information about the kernel 202, such as by identifying where the first instruction set and / or associated information is stored and / or identifying the characteristics or other information of the buffer 104.

[0020] In some examples, the information of kernel 202 generated by computing system 100 and stored in trace data structure 108 includes or otherwise indicates: one or more storage locations of a first instruction set and / or associated information corresponding to kernel 202 in buffer 104, one or more semaphore values ​​associated with the first instruction set and / or associated information (e.g., to be released upon instruction completion or execution), one or more addresses of semaphores associated with the first instruction set and / or associated information, one or more channels and / or variations thereof associated with the first instruction set and / or associated information. In some embodiments, the information of kernel 202 generated by computing system 100 and stored in trace data structure 108 includes or otherwise indicates one or more locations or entries in buffer 104 that store or otherwise indicate the first instruction set and / or associated information. A storage location, such as the storage location described herein (which may also be referred to as a storage address, memory address, memory location, and / or variations thereof), may refer to a location in the memory or storage device of one or more computing devices and may be indicated in any suitable manner, such as a memory address, a reference to a memory address, a virtual address, a range of memory addresses, a location in a data structure, an index, and / or variations thereof.

[0021] Figure 3 Example 300 is illustrated as a portion of a process for monitoring submitted tasks according to at least one embodiment. Example 300 may be a continuation of the process of Example 200. In one embodiment, computing system 100 monitors tasks submitted by one or more host processing units 102 to one or more device processing units 106 using buffer 104. User programs may instruct one or more kernels to be submitted by one or more host processing units 102 for execution by one or more device processing units 106. (See attached...) Figure 2 As indicated, computing system 100 may be able to track or otherwise monitor tasks enqueued into buffer 104 as a result of invoking the start monitor API. Computing system 100 may execute the start monitor API to enable computing system 100 to begin tracking or otherwise monitoring tasks enqueued into buffer 104 by storing at least information about the enqueued tasks in the tracking data structure 108.

[0022] The user program can instruct kernel 302, which may be a kernel to be executed by one or more device processing units 106. As part of execution by one or more host processing units 102, the user program can cause one or more host processing units 102 to submit kernel 302 for execution by one or more device processing units 106, wherein kernel 302 may first be enqueued or otherwise stored in buffer 104. In one embodiment, kernel 302 is enqueued, instructed, or otherwise stored in buffer 104 as a second instruction set and / or associated information, which can be implemented using any suitable data or format. As a result of executing the start monitor API and submitting kernel 302 for execution by one or more device processing units 106, computing system 100 can generate information about kernel 302 and store this information in trace data structure 108.

[0023] In one example, the information of kernel 302 generated by computing system 100 and stored in trace data structure 108 includes, or otherwise indicates, one or more storage locations of a second instruction set and / or associated information corresponding to kernel 302 in buffer 104, one or more semaphore values ​​associated with the second instruction set and / or associated information (e.g., to be released upon instruction completion or execution), one or more addresses of semaphores associated with the second instruction set and / or associated information, one or more channels and / or variations thereof associated with the second instruction set and / or associated information. The information of kernel 302 generated by computing system 100 and stored in trace data structure 108 may include, or otherwise indicate, one or more locations or entries in buffer 104 that store or otherwise indicate the second instruction set and / or associated information.

[0024] Figure 4 Example 400 is illustrated as a portion of a process for monitoring submitted tasks according to at least one embodiment. Example 400 may be a continuation of the process of Example 300. In one embodiment, computing system 100 monitors tasks submitted by one or more host processing units 102 to one or more device processing units 106 using buffer 104. User programs may instruct one or more kernels to be submitted by one or more host processing units 102 for execution by one or more device processing units 106. (See also...) Figure 2 As indicated, this allows computing system 100 to track or otherwise monitor tasks that are queued into buffer 104 as a result of invoking the start monitor API.

[0025] The computing system 100 can execute a start monitor API to cause the computing system 100 to begin tracking or otherwise monitor tasks enqueued into buffer 104 by storing at least information about the enqueued tasks in the tracking data structure 108. The computing system 100 can track or otherwise monitor all tasks submitted for execution by one or more device processing units 106 until a stop monitor API is invoked. The user program may include invoking the start monitor API, submitting one or more tasks for execution by one or more device processing units 106, and invoking the stop monitor API, wherein the computing system 100 can track or otherwise monitor one or more tasks by storing information about one or more tasks in the tracking data structure 108. The user program may then include invoking the stop monitor API to cause the computing system 100 to stop tracking or otherwise monitor tasks enqueued into buffer 104. More information about the stop monitor API is available at [link to relevant documentation]. Figure 6 It was found in the description.

[0026] After invoking the Stop Monitor API, the user program can instruct kernel 402, which can be a kernel to be executed by one or more device processing units 106. The user program can cause one or more host processing units 102 to submit kernel 402 for execution by one or more device processing units 106, wherein kernel 402 can be enqueued, instructed, or otherwise stored in buffer 104 as third instruction set and / or associated information, which can be implemented using any suitable data or format. As a result of the Stop Monitor API, computing system 100 may not generate information about kernel 402 and may not store such information in trace data structure 108 (e.g., the tracing process of computing system 100 is no longer active). In one embodiment, because computing system 100 has executed the Stop Monitor API to cause computing system 100 to stop storing or otherwise instructing information about tasks enqueued in buffer 104, computing system 100 does not generate information about kernel 402 and does not store such information in trace data structure 108.

[0027] Figure 5An example 500 of an API for initiating a monitoring task is illustrated according to at least one embodiment. In one embodiment, a user program utilizes, invokes, or otherwise calls a start monitor API 502 to cause a computing system to begin tracking or other monitoring of queued tasks. The computing system may include one or more processors (e.g., a host processing unit and / or a device processing unit) having one or more circuits that execute the start monitor API 502, such as by executing one or more instructions of the start monitor API 502. The computing system may be combined with... Figure 1-4 The described computing system is consistent.

[0028] The user program may instruct a task (e.g., one or more kernels) to be executed by the device processing unit. In one embodiment, the user program includes: invoking a start monitor API 502 to cause the computing system to begin tracing or otherwise monitoring the submitted task, and submitting the task for execution by the device processing unit. As part of the execution, the user program may cause the host processing unit to submit the task for execution by the device processing unit, which may cause the generation of one or more instructions and / or other information associated with the task and the storage of such instructions and / or other information in a buffer such as the buffer described herein, wherein the one or more instructions and / or other information can be scheduled for execution by the device processing unit, thereby executing the task. In one embodiment, the start monitor API 502 causes the computing system to store or otherwise indicate information about the submitted task in a trace data structure such as the trace data structure described herein.

[0029] In at least one embodiment, the tracking data structure is defined by the following code, but any variation thereof can be used: / * Batch of work * / struct workTracker { Bool is Monitored; U32 recorded CTS; channel *channel; trackedNodes *trackedNodesHead; workTracker *next, *prev; }; / * Each launch's tracked information * / struct trackedNodes { Current *CurrentStart, *CurrentEnd; U64 qmdSemaOffset; U32 qmdSemaPayload; channel *channel; trackedNodes *next, *prev; };” In one embodiment, the Start Monitor API 502 (also referred to as the Stream Start Monitor API, Start Monitor Task API, Start Monitor Submitted Task API, and / or variations thereof) receives stream 504 as an input parameter. In some examples, the Start Monitor API 502 may receive any suitable input parameter in addition to or in addition to the parameters described herein. In at least one embodiment, stream 504 is any suitable identifier, indication, or reference to a stream. A stream can refer to a sequence of operations to be performed by a device processing unit, wherein tasks can be submitted for execution by a device processing unit by indicating or otherwise posting tasks in the stream. Streams such as those described herein can refer to any suitable work submission mechanism, system, or process that can be used to cause one or more processing units (e.g., device processing units) to perform operations or otherwise provide data to one or more processing units for the performance of operations.

[0030] As an illustrative example, a computing system performs the Start Monitor API 502 by at least the following actions: generating or otherwise obtaining a workTracker trackObj and adding it to the header of stream->trackObjPool; setting trackObj->isMonitored = TRUE; and / or setting trackObj->channel=channel and trackObj->recordedCTS to the value of the channel associated with the given stream (e.g., the value of a tracking semaphore, such as the value last emitted by the CPU). For each kernel launched in this stream, if trackObjPool->trackObj.isMonitored == TRUE, the computing system can spawn a new node and add that new node to trackObj->trackedNodesHead; before writing the QMD method to the push buffer (PB), node->CurrentBegin is populated with Current, and after writing the QMD method to the PB, node->CurrentEnd is populated with Current; node->qmdSemaPayload is populated with a value to be released upon launch (e.g., a value associated with the final payload of the semaphore); and / or node->qmdSemaOffset is populated with the virtual address (VA) of the semaphore (e.g., semaphoreGetOffset(qmd->semaphore)). In at least one embodiment, the QMD method refers to any suitable scheduled instruction, command, message, method, and / or variant thereof that can be used by the device processing unit to execute a queued task. Additionally, it should be noted that the names, identifiers, and / or references to variables, data, information, functions, and / or procedures (such as those described herein) are intended to be illustrative examples, and such names, identifiers, and / or references may take any suitable syntax or format and may reference any suitable programming model and / or functionality or purpose. As an illustrative example, a structure such as the “workTracker” described herein may be referenced in any suitable manner (such as a tracker data structure, a trace data structure, a work trace structure, and / or variations thereof) and may reference any suitable programming model (such as CUDA, HIP, oneAPI, and / or variations thereof). Similar variations may be applied to any names, identifiers, and / or references described herein.

[0031] The computing system can execute the Start Monitor API 502 at least by: after a user program invokes the Start Monitor API 502, for any task submitted by the host processing unit to be executed by the device processing unit using a stream, when the task is queued for execution by the device processing unit at least by storing instructions for executing the task in a buffer, causing the computing system to generate information about the stored instructions and store that information in a trace data structure, such as the storage location of the stored instructions in the buffer (e.g., start and end addresses), the channel associated with the stored instructions (e.g., one or more channels associated with the stored instructions and / or the buffer), and the semaphore value and address associated with the execution of the stored instructions. The computing system can indicate the information to one or more users by storing the information in the trace data structure, which can be accessed by one or more users. In one embodiment, in response to the Start Monitor API 502, the computing system generates and provides a Start Monitor API return 506. In at least one embodiment, the return is provided without being combined with the Start Monitor API 502. The Start Monitor API return 506 may include a status 508, which can be any suitable indication or data regarding the execution status of the Start Monitor API 502, such as whether a process, like the one described herein, succeeded, failed, or encountered other errors. The trace data structure can be used by one or more users to locate stored instructions, such as to modify the stored instructions, thereby modifying or preventing the device processing unit from performing tasks.

[0032] Figure 6 An example 600 of an API for stopping monitoring tasks is illustrated according to at least one embodiment. In one embodiment, a user program utilizes, invokes, or otherwise calls a stop monitor API 602 to cause a computing system to stop tracking or otherwise monitoring queued tasks. The computing system may include one or more processors (e.g., a host processing unit and / or a device processing unit) having one or more circuits for executing the stop monitor API 602, such as by executing one or more instructions of the stop monitor API 602. The computing system may be combined with... Figure 1-5 The described computing system is consistent.

[0033] In one embodiment, the user program includes at least: invoking a function to cause the computing system to begin tracking or otherwise monitoring the submitted task. Figure 5 The start monitor API 502 and the stop monitor API 602 are invoked to cause the computing system to stop tracking or otherwise monitor the submitted task. In one embodiment, if the computing system was previously caused to start tracking or otherwise monitor the submitted task (e.g., as a stop monitor API 602), the stop monitor API 602 is invoked. Figure 5 If the result of the start monitor API 502 is obtained, then the stop monitor API 602 causes the computing system to stop tracking or otherwise monitoring the submitted task in a trace data structure such as that described herein.

[0034] In one embodiment, the Stop Monitor API 602 (also referred to as the Stream Stop Monitor API, Stop Monitor Task API, Stop Monitor Submitted Task API, and / or variations thereof) receives a stream 604 and a trace object 606 as input parameters. In some examples, the Stop Monitor API 602 may receive any suitable input parameters in lieu of or in addition to those described herein. In at least one embodiment, the stream 604 is any suitable identifier, indication, or reference to a stream. In at least one embodiment, the trace object 606 is any suitable identifier, indication, or reference to a trace data structure. The trace data structure may be generated as part of a user program and may be used to store information about tasks submitted using a stream (e.g., as...). Figure 5 (The results of the start monitor API 502).

[0035] As an illustrative example, the computing system executes Stop Monitor API 602 at least by setting trackObjPool->trackObj.isMonitored = FALSE and returning trackObj. The computing system may execute Stop Monitor API 602 at least by ensuring that, after invoking Stop Monitor API 602, the computing system does not generate or store any information about any subsequent tasks submitted by the host processing unit for execution by the device processing unit using the stream. In some examples, the computing system executes Stop Monitor API 602 at least by ensuring that, after invoking Stop Monitor API 602, the computing system does not generate any information associated with any subsequent queued tasks to be executed by the device processing unit and / or does not store that information in the tracking data structure.

[0036] The computing system may execute the Stop Monitor API 602 at least by causing the computing system not to store, or no longer store, any new information corresponding to a new instruction stored in a buffer as the result of a task submitted by a host processing unit for execution by a device processing unit in the trace data structure. The computing system may also execute the Stop Monitor API 602 at least by preventing the computing system from storing any new information corresponding to a new instruction stored in a buffer as the result of a task submitted by a host processing unit for execution by a device processing unit in the trace data structure, which may prevent such information from being indicated to one or more users via the trace data structure. In one embodiment, in response to the Stop Monitor API 602, the computing system generates and provides a Stop Monitor API return 608. In at least one embodiment, the return is provided without being combined with the Stop Monitor API 602. The Stop Monitor API return 608 may include a state 610, which may be any suitable indication or data of the execution status of the Stop Monitor API 602, such as whether a process, like the process described herein, succeeded, failed, or encountered other errors. The Stop Monitor API return 608 may include the trace data structure or other references or indications to the trace data structure.

[0037] Figure 7 An example of a process 700 indicating the location of a scheduled instruction according to at least one embodiment is illustrated. In one embodiment, one or more portions of process 700 are composed of, as in combination with Figure 1-6 The described computing system executes the process. In some embodiments, some or all of the processes in process 700 (or any other process described herein, or variations and / or combinations thereof) are executed under the control of one or more computer systems configured with computer-executable instructions and implemented by hardware, software, or a combination thereof as code (e.g., computer-executable instructions, one or more computer programs, or one or more application programs) that execute jointly on one or more processors. One or more portions of process 700 may be executed in any suitable order, including sequential, parallel, and / or variations thereof, and while process 700 may be depicted as a series of steps or operations, process 700 may include modified or reordered steps or operations, or omit certain steps or operations, unless explicitly stated or logically required, such as when the output of one step or operation is used as the input of another step or operation.

[0038] At box 702, the computing system can execute the API to indicate the storage location of the scheduled instruction. The computing system can obtain a user program, which can be represented as any suitable computer program code, and this computer program code can include calls to the API, which can be... Figure 5The start monitor API 502. Calls to the API can include instructions for streams. As part of executing user programs, the computing system can execute... Figure 5 The Start Monitor API 502. As part of executing a user program, the computing system may receive an indication, such as an API described herein, that causes one or more storage locations of one or more scheduled instructions to be indicated to one or more users. This indication may be an API function or call (e.g., the Start Monitor API as indicated in the computer program code) and may cause the computing system to execute the API as described herein. In one embodiment, as a result of executing the API, for any task submitted by the host processing unit to be executed by the device processing unit using the indicated stream, when the task is enqueued for execution by the device processing unit at least by storing instructions for executing the task in a buffer, the computing system generates information about the stored instructions (such as the storage location of the stored instructions in the buffer and the semaphore values ​​and addresses associated with the execution of the stored instructions) and stores this information in a trace data structure.

[0039] At block 704, the computing system can determine whether any scheduled instructions exist. In one embodiment, a scheduled instruction refers to any suitable information that can be used to execute a task enqueued for execution. As an illustrative example, a user program instructs a kernel to be executed by a device processing unit, wherein the host processing unit stores instructions for executing the kernel in a buffer, wherein the stored instructions may be referred to as scheduled instructions and may indicate aspects for executing the kernel. Continuing this example, the device processing unit obtains the instructions in the buffer for executing the kernel. The computing system can determine whether any scheduled instructions exist by, as part of executing the user program, determining whether any tasks are being submitted for execution by the device processing unit using the indicated stream and / or whether any instructions are stored in a buffer for the device processing unit to schedule for task execution, these instructions may indicate (e.g., task-associated) scheduled instructions. If so, the computing system proceeds to block 706 and stores information about the scheduled instructions (such as the location of the scheduled instructions in the buffer, the channel associated with the buffer and / or the scheduled instructions, and the semaphore value and address associated with the execution of the scheduled instructions) in a trace data structure. In some examples, the location of the scheduled instruction can be indicated by any suitable indication of the start and end addresses, index, entry, and / or location in the storage device or memory. If no scheduled instruction exists, the computing system can proceed to block 708.

[0040] At box 708, the system can determine whether to stop storing the location. This is retrieved by the user program. Figure 6The computing system can determine a stop storage location by using the Stop Monitor API 602 or by having a user program otherwise cause the computing system to execute the result of the Stop Monitor API 602. As part of executing the user program, the computing system can receive an indication to block the storage of one or more scheduled instructions at one or more storage locations via an API such as the API described herein. This indication can be an API function or call (e.g., a Stop Monitor API as indicated in the computer program code) and can cause the computing system to execute the API as described herein. This indication allows the system to determine the stop storage location. If, as part of executing the user program, the user program calls... Figure 6 If the stop monitor API 602 is invoked, the computing system can determine the stop storage location. If the computing system determines there is no stop storage location, the computing system proceeds to box 704.

[0041] If the computing system determines a stopping storage location, the computing system proceeds to box 710 to execute an API to prevent the storage location of the scheduled instruction from being stored. This API can be... Figure 6 The stop monitor API 602. As part of executing user programs, the computing system can execute... Figure 6 The stop monitor API 602. A computing system may execute the API, at least by causing the computing system to prevent information such as the storage location of any subsequent scheduled instructions from being indicated or otherwise stored in a trace data structure. The computing system may obtain indications of subsequent scheduled instructions (e.g., based on computer program code) and (e.g., by preventing one or more processes of storing the storage location of the scheduled instructions from occurring or by otherwise providing one or more indications that cause the computing system not to store any storage location of the scheduled instructions) prevent the storage location of the scheduled instructions from being stored in the trace data structure.

[0042] Data Center Figure 8An example data center 800 according to at least one embodiment is illustrated. The data center 800 may include one or more rooms having racks 802 and auxiliary equipment for housing one or more racks 802 and one or more substrates 804. Racks 802 may include one or more substrates 804. Racks 802 may include housings for housing and supporting individual substrates 804. Operational aspects of racks 802 may be adjustable at the rack level (corresponding to a group of substrates 804) or at the substrate level (corresponding to an individual substrate 804), among other options. Racks 802 or substrates 804 may have specific selected maximum operating parameters, such as, but not limited to, power consumption, operating frequency, etc. The data center 800 may be supported by various cooling systems, such as, but not limited to, cooling towers, cooling loops, pumps, and other support systems. The cooling system may include sensors and controllers for monitoring and managing the cooling characteristics of racks 802. The substrates 804 within racks 802 may draw operating power from one or more power distribution units (PDUs; not shown). PDUs can be arranged within racks 802, for example, between racks 802 that include substrates 804, or within racks 802 that also house substrates 804.

[0043] Rack 802 and substrate 804 may include subsystems, modules, add-in cards, and other semiconductor components. Substrate 804 may include one or more computing units 806, each computing unit 806 may include one or more processors 808, one or more memories 810, and an interface controller 812. Computing unit 806 may include any number of processors, such as, but not limited to, a central processing unit (“CPU”), a graphics processing unit (“GPU”), or other processors (including accelerators, field-programmable gate arrays (FPGAs), graphics processors, etc.), including any processors described herein, such as, but not limited to, those described herein. Figure 9-21B The processor in the computing unit 806 may include one or more memory storage devices 810 (e.g., dynamic read-only memory, solid-state storage devices, or disk drives), as well as network input / output (“NW I / O”) devices, network switches, virtual machines (“VMs”), power supply modules, and cooling modules, etc. One or more computing units 806 may be a server having one or more of the aforementioned computing resources.

[0044] Computing unit 806 may include individual computing unit groups housed in one or more racks (not shown), or in numerous racks within data centers in different geographical locations (also not shown). Individual computing unit groups may include grouped computing, networking, memory, or storage resources that can be configured or allocated to support one or more workloads. Several computing units (e.g., including CPUs and / or other processors) may be grouped within one or more racks to provide computing resources to support one or more workloads. Resource coordinator 814 may configure or otherwise control one or more computing units 806 or groups of computing units. Resource coordinator 814 may include a Software Design Infrastructure (“SDI”) management entity for data center 800. Resource coordinator 814 may include hardware, software, or some combination thereof.

[0045] Data center 800 may include any one or any combination of the framework layer 820, software layer 830, and application layer 840. For example... Figure 8 As shown, framework layer 820 includes job scheduler 822, configuration manager 824, resource manager 826, and distributed file system 828. Framework layer 820 may include frameworks for supporting software 832 of software layer 830 and / or one or more applications 842 of application layer 840. Software 832 or application 842 may respectively include web-based service software or applications, such as, but not limited to, software or applications provided by Amazon Web Services, Google Cloud, and Microsoft Azure. Framework layer 820 may be a type of free and open-source software web application framework, such as, but not limited to, Apache Spark™ (hereinafter “Spark”), which can leverage distributed file system 828 for large-scale data processing (e.g., “big data”). Job scheduler 822 may include Spark drivers, which facilitate the scheduling of workloads supported by various layers of data center 800. Configuration manager 824 may be able to configure different layers, such as, but not limited to, software layer 830 and framework layer 820 (which includes Spark and distributed file system 828 for supporting large-scale data processing). Resource manager 826 can manage clustered or grouped computing units 806 that are mapped to or allocated to support distributed file system 828 and job scheduler 822. Resource manager 826 can coordinate with resource coordinator 814 to manage these mapped or allocated computing resources.

[0046] Software 832 may be included in software layer 830, and may include software used by at least a portion of computing units 806, one or more computing units 806, groups of computing units 806, and / or the distributed file system 828 of framework layer 820. One or more types of software may include, but are not limited to, internet web search software, email virus scanning software, database software, and streaming video content software.

[0047] Application 842 may be included in application layer 840 and may include one or more types of applications used by at least the portions of computing unit 806, one or more computing units 806, groups of computing units 806, and / or the distributed file system 828 of framework layer 820. One or more types of applications may include, but are not limited to, any number of genomics applications, cognitive computing applications, and machine learning applications, including training or inference software, machine learning framework software (e.g., PyTorch, TensorFlow, Caffe, etc.), or other machine learning applications used in conjunction with one or more embodiments.

[0048] Any of the Configuration Manager 824, Resource Manager 826, and Resource Coordinator 814 can implement any number and type of self-modification actions based on any amount and type of data obtained in any technically feasible manner. Self-modification actions can alleviate the burden on data center operators of Data Center 800 to make potentially erroneous configuration decisions and may prevent underutilized and / or poorly performing portions of the data center.

[0049] Data center 800 may include tools, services, software, or other resources for training one or more machine learning models according to one or more embodiments described herein, or for using one or more machine learning models to predict or infer information. For example, a machine learning model can be trained by calculating weight parameters based on a neural network architecture using the software and computing resources described above regarding data center 800. The trained machine learning model corresponding to one or more neural networks can be used with the resources described above regarding data center 800 to infer or predict information using weight parameters calculated through one or more training techniques described herein.

[0050] Data Center 800 can use CPUs, application-specific integrated circuits (ASICs), GPUs, FPGAs, or other hardware (e.g., Figure 9-21BThe embodiments described herein can be used to perform some or all of the processes and techniques described elsewhere, such as, but not limited to, training and / or inference using the resources described above. Furthermore, one or more of the software and / or hardware resources described above can be configured as a service to allow a user to train or perform information inference, such as, but not limited to, image recognition, speech recognition, or other artificial intelligence services.

[0051] In at least one embodiment, processor 808 may include one of the processors described below and / or include one or more circuitry configured to: execute an API to cause one or more storage locations of one or more scheduled instructions to be indicated to one or more users; execute an API to prevent one or more storage locations of one or more scheduled instructions from being stored; and / or execute statements regarding... Figure 1-7 One or more of the processes illustrated or described, or otherwise performing any of the operations described above or elsewhere herein. In at least one embodiment, processor 808 is configured by software 832 to: execute an API to cause one or more storage locations of one or more scheduled instructions to be indicated to one or more users; execute an API to prevent one or more storage locations of one or more scheduled instructions from being stored; and / or perform operations related to... Figure 1-7 One or more of the processes illustrated or described, or any of the operations described above or elsewhere herein, may be performed. Data Center 800 may use logic, CPU, application-specific integrated circuits (ASICs), GPUs, FPGAs, or other hardware (e.g., Figure 9-21B (In the embodiments) perform any of the operations described above or elsewhere in this document.

[0052] processor The following figures illustrate, but are not limited to, an example processor and processing system that can be used to: execute an API to cause one or more storage locations of one or more scheduled instructions to be indicated to one or more users; execute an API to prevent one or more storage locations of one or more scheduled instructions from being stored; and / or perform actions related to... Figure 1-7 One or more of the processes illustrated or described herein, or otherwise performing some or all of the processes, operations, and / or techniques described elsewhere herein. Example processors and processing systems may be software-configurable to: execute APIs to cause one or more storage locations of one or more scheduled instructions to be indicated to one or more users; execute APIs to prevent one or more storage locations of one or more scheduled instructions from being stored; and / or perform operations related to... Figure 1-7One or more of the processes illustrated or described, or any operation performed above or elsewhere herein, may be included. Processors and processing systems may include logic, a central processing unit (CPU), an application-specific integrated circuit (ASIC), a graphics processing unit (GPU), a field-programmable gate array (FPGA), an XPU (i.e., any computing architecture best suited to the needs of the application), or other hardware (e.g., Figure 9-21B The embodiments described herein are used to perform any of the operations described above, below, or elsewhere herein. The processor and / or processing system described herein may include one or more circuitry that can be used to execute an API to indicate one or more storage locations of one or more scheduled instructions to one or more users; execute an API to prevent one or more storage locations of one or more scheduled instructions from being stored; and / or execute statements regarding... Figure 1-7 One or more of the processes illustrated or described herein, or any of the operations described above or elsewhere herein. As used herein, one or more circuits may be configured by software to execute APIs to cause one or more storage locations of one or more scheduled instructions to be indicated to one or more users; execute APIs to prevent one or more storage locations of one or more scheduled instructions from being stored; and / or perform operations related to... Figure 1-7 One or more of the processes illustrated or described, or any of the operations performed above or elsewhere in this document. Figure 26A and Figure 26B The illustration depicts logic 2615 according to at least one embodiment, which, as described elsewhere herein, can be used in one or more devices to perform operations such as, but not limited to, those discussed herein. For example, logic can refer to any combination of software logic, hardware logic, and / or firmware logic that provides the functionality and / or operations described herein, wherein the logic can be collectively or individually embodied as part of a circuit system forming a larger system, such as an integrated circuit (IC), application-specific integrated circuit (ASIC), field-programmable gate array (FPGA), system-on-a-chip (SoC), or one or more processors (e.g., CPU, GPU).

[0053] Figure 9A processor according to at least one embodiment is illustrated, which is a system-on-a-chip (SOC) 900 (may be referred to as a system-on-a-chip, superchip, or other names). The SOC 900 may include processor complexes 910 and 940. The SOC 900 may include any number of processor complexes 910 and / or processor complexes 940, which may include any number of processors described herein in any combination, such as, but not limited to, processors described herein in any combination. Figure 9-21B The processor in the system. For example, processor 910 may include a central processing unit (CPU), and processor 940 may include a graphics processor. Alternatively, processor 910 may include a graphics processor, and processor 940 may include a graphics processor. The SOC 900 may include any number of display controllers 992, any number of multimedia engines 994, any number of I / O interfaces 970, any number of memory controllers 980, and any number of fabrics 960 in any combination. For ease of explanation, this document uses reference numbers identifying objects and bracket numbers identifying instances (if necessary) to denote multiple instances of similar objects. The SOC 900 may include a processor from Broadcom Corporation, Palo Alto, California.

[0054] Processor complex 910 may include a CPU, processor complex 940 may include a GPU, and SOC 900 may include a processing unit integrating processor complex 910 and processor complex 940 onto a single chip. Certain tasks may be assigned to processor complex 910, while other tasks may be assigned to processor complex 940. Processor complex 910 may be configured to execute main control software associated with SOC 900, such as, but not limited to, an operating system. Processor complex 910 may be the main processor of SOC 900, controlling and coordinating the operation of other processors. Processor complex 910 may issue commands that control the operation of processor complex 940 to perform some or all of the operations described herein. Processor complex 910 may be configured to execute host-executable code derived from CUDA or other source code (e.g., HIP source code), while processor complex 940 may be configured to execute device-executable code derived from CUDA or other source code to perform any of the operations described herein.

[0055] The processor complex 910 may include cores 920(1)-920(4) and cache (e.g., L3 cache) 930 for storing information for performing the operations described herein. The processor complex 910 may include any number of cores 920 in any combination and any number and type of cache. The cores 920 may be configured to execute instructions of a specific instruction set architecture (“ISA”) to perform some or all of the operations described herein. Each core 920 may include a CPU core. Cores 920(1)-920(4) may be referred to as compute units or arithmetic units. The SOC 900 may include any number of processor complexes 910, architecture 960, I / O interface 970, and memory controller 980.

[0056] Each core 920 may include a fetch / decode unit 922, an integer execution engine 924, a floating-point execution engine 926, and an L2 cache 928. The fetch / decode unit 922 may fetch instructions to perform some or all of the operations described herein (e.g., but not limited to APIs compiled into instructions) and decode those instructions, generate micro-operations, and dispatch individual micro-instructions to the integer execution engine 924 and / or the floating-point execution engine 926. The fetch / decode unit 922 may concurrently dispatch one micro-instruction to the integer execution engine 924 and another micro-instruction to the floating-point execution engine 926. The integer execution engine 924 may perform integer and memory operations. The floating-point engine 926 may perform floating-point and vector operations. The fetch / decode unit 922 may dispatch micro-instructions to one or more execution engines, which may replace both the integer execution engine 924 and the floating-point execution engine 926.

[0057] Each core 920(i) (where i is an integer representing a specific instance of core 920) can access the L2 cache 928(i) included in core 920(i). Each core 920 included in core complex 910(j) (where j is an integer representing a specific instance of core complex 910) can be connected to other cores 920 included in core complex 910(j) via the L3 cache 930(j) included in core complex 910(j). The cores 920 included in core complex 910(j) (where j is an integer representing a specific instance of core complex 910) can access all L3 caches 930(j) included in core complex 910(j). The L3 cache 930 can include any number of slices.

[0058] Processor complex 940 may be a graphics complex that can be configured to perform computational operations (e.g., the computational operations described herein) in a highly parallel manner. Processor complex 940 may be configured to perform graphics pipeline operations, such as, but not limited to, drawing commands, pixel operations, geometric calculations, and other operations associated with rendering an image to a display. Processor complex 940 may be configured to perform graphics-independent operations, such as, but not limited to, neural network training and / or simulation. Processor complex 940 may be configured to perform both graphics-related and graphics-independent operations.

[0059] The processor complex 940 may include any number of compute units 950(1)-950(N) (where N is any integer greater than 1) and an L2 cache 942. The compute units 950 may share the L2 cache 942, which may store information that will be used to perform some or all of the operations described herein. The L2 cache 942 may be partitioned. The processor complex 940 may include any number of compute units 950 and any number and type of cache. The processor complex 940 may include any number of dedicated graphics hardware.

[0060] Each compute unit 950 may include any number of SIMD units 952(1)-952(N) (where N is any integer greater than 1) and shared memory 954. Each SIMD unit 952 may implement a SIMD architecture and may be configured to perform some or all of the operations described herein in parallel. Each compute unit 950 may execute any number of thread blocks, but each thread block may execute on a single compute unit 950, although in some embodiments, the thread block may execute on multiple compute units. A thread block may include any number of execution threads. A workgroup may be a thread block. Each SIMD unit 952 may execute a set of threads. A set of threads (e.g., 16 threads), also referred to as a warp, subgroup, or wavefront (e.g., used by AMD and Intel), may belong to a single thread block and be configured to process different datasets based on a single instruction set. Prediction may be used to disable one or more threads in a warp, subgroup, or wavefront. A lane may be a thread. A work item can be a thread, such as (but not limited to) an OpenCL thread. Different thread bundles, subgroups, or wavefronts within a thread block can be synchronized together and communicate via shared memory 954. Each compute unit 950 can include one or more thread block clusters, where thread block clusters can implement programmable control over locality at a larger granularity than a single thread block in a single streaming multiprocessor (SM). Thread block clusters (also referred to as “clusters”) can support multiple thread blocks running concurrently across streaming multiprocessors, thereby synchronously and cooperatively acquiring, exchanging, or otherwise using data. In at least one embodiment, a streaming multiprocessor (“SM”) can refer to a streaming microprocessor, a streaming processor (“SP”), a streaming processing unit (“SPU”), a compute unit (“CU”), an execution unit (“EU”), and / or a slice, where a slice in this context can refer to a portion of the processing resources within a processing unit (e.g., 16 cores, a ray tracing unit, a thread bootstrap, or a scheduler).

[0061] Structure 960 may be a system interconnect that facilitates data and control transfers across processor complex 910, processor complex 940, I / O interface 970, memory controller 980, display controller 992, and multimedia engine 994, for example, to perform some or all of the operations described herein. SOC 900 may include any number and type of system interconnects other than or replacing structure 960, facilitating data and control transfers across any number and type of directly or indirectly linked components within or outside SOC 900. I / O interface 970 may represent any number and type of I / O interfaces (e.g., PCI, PCI extensions (“PCI-X”), PCIe, Gigabit Ethernet (“GBE”), USB, etc.). Various types of peripheral devices may be coupled to I / O interface 970. Peripherals that may be coupled to I / O interface 970 may include keyboards, mice, printers, scanners, joysticks or other types of game controllers, media recording devices, external storage devices, network interface cards, etc.

[0062] Display controller 992 can display images on one or more display devices, such as, but not limited to, liquid crystal displays (“LCD” devices. Multimedia engine 994 can include any number and type of circuitry related to multimedia, such as, but not limited to, video decoders, video encoders, image signal processors, etc. Memory controller 980 can facilitate data transfer between SOC 900 and unified system memory 990. Processor complex 910 and processor complex 940 can share unified system memory 990. Unified system memory 990 can include various types of memory devices, including dynamic random access memory (DRAM) or graphics random access memory, such as, but not limited to, synchronous graphics random access memory (SGRAM), including graphics double data rate (GDDR) memory. Unified system memory 990 can include 3D stacked memory, including but not limited to high bandwidth memory (HBM), HBM2e, or HDM3.

[0063] The SOC 900 can implement a memory subsystem comprising any number and type of memory controllers 980 and memory devices (e.g., shared memory 954), which may be dedicated to a single component or shared among multiple components to perform any of the operations described herein. The SOC 900 can implement a cache subsystem comprising one or more cache memories (e.g., L2 cache 928, L3 cache 930, and L2 cache 942), each cache memory may be dedicated to any number of components (e.g., core 920, core complex 910, SIMD unit 952, compute unit 950, and processor complex 940) or shared among any number of components (e.g., core 920, core complex 910, SIMD unit 952, compute unit 950, and processor complex 940).

[0064] In at least one embodiment, the SOC 900 may include one or more circuits for executing an API to cause one or more storage locations of one or more scheduled instructions to be indicated to one or more users; executing an API to prevent one or more storage locations of one or more scheduled instructions from being stored; and / or executing regarding Figure 1-7 One or more of the processes illustrated or described, or otherwise performing any of the operations described above or elsewhere herein. One or more circuits may be software-configured to execute APIs to cause one or more storage locations of one or more scheduled instructions to be indicated to one or more users; to execute APIs to prevent one or more storage locations of one or more scheduled instructions from being stored; and / or to perform operations related to... Figure 1-7 One or more of the processes illustrated or described, or any of the operations performed above or elsewhere in this document.

[0065] Figure 10A A parallel processor 1000 according to at least one embodiment is illustrated. The parallel processor 1000 can be implemented using one or more circuits and can be referred to as a programmable processor (e.g., CPU and / or GPU), logic, application-specific integrated circuit (ASIC), field-programmable gate array (FPGA), or other hardware (e.g., Figure 9-21B (The embodiments in the document) are used to perform any of the operations described above or elsewhere herein.

[0066] Parallel processor 1000 may include parallel processing unit 1002 for performing any of the operations described above or elsewhere herein. Parallel processing unit 1002 may include I / O unit 1004 that enables communication with other devices, including other instances of parallel processing unit 1002. I / O unit 1004 may be directly connected to other devices. I / O unit 1004 may be connected to other devices via the use of a hub or switch interface, such as, but not limited to, memory hub 1005. The connection between memory hub 1005 and I / O unit 1004 may form a communication link 1013. I / O unit 1004 may be connected to host interface 1006 and memory crossbar switch 1016, wherein host interface 1006 receives commands directed to perform processing operations, and memory crossbar switch 1016 receives commands directed to perform memory operations.

[0067] When host interface 1006 receives a command buffer via I / O unit 1004, host interface 1006 can route the work operations that execute these commands to front-end 1008. Front-end 1008 can be coupled to scheduler 1010 (which may be referred to as sequencer), which is configured to distribute commands or other work items to processing cluster array 1012. Scheduler 1010 can ensure that processing cluster array 1012 is correctly configured and in an active state before tasks are distributed to the cluster of processing cluster array 1012. Scheduler 1010 can be implemented via firmware logic executed on a microcontroller. The microcontroller-implemented scheduler 1010 can be configured to perform complex scheduling and work distribution operations at both coarse and fine granular levels, thereby enabling fast preemption and context switching of threads executing on processing array 1012. Host software can validate workloads scheduled on processing cluster array 1012 via one of multiple graphics processing paths. The workload can then be automatically distributed to the processing array cluster 1012 by the scheduler 1010 logic within the microcontroller, which includes the scheduler 1010.

[0068] Processing cluster array 1012 can perform any of the operations described above or elsewhere herein, and may include up to “N” processing clusters (e.g., clusters 1014A, 1014B through 1014N), where “N” represents a positive integer (which may be a different integer “N” than used in other diagrams). Each cluster 1014A-1014N in processing cluster array 1012 can execute a large number of concurrent threads. Scheduler 1010 may use various scheduling and / or work distribution algorithms to distribute work to clusters 1014A-1014N in processing cluster array 1012, which may vary depending on the workload generated by each type of program or computation. Scheduling may be dynamically handled by scheduler 1010 or may be assisted by compiler logic during the compilation of program logic configured to be executed by processing cluster array 1012. Different clusters 1014A-1014N of processing cluster array 1012 may be assigned to process different types of programs or perform different types of computations.

[0069] The processing cluster array 1012 can be configured to perform various types of parallel processing operations, such as, but not limited to, any of the operations described above or elsewhere herein. The processing cluster array 1012 can be configured to perform general-purpose parallel computing operations. For example, the processing cluster array 1012 may include logic for performing processing tasks, including filtering video and / or audio data, performing modeling operations (including physical operations), and performing data transformations.

[0070] Processing cluster array 1012 can be configured to perform parallel graphics processing operations. Processing cluster array 1012 may include additional logic for supporting the execution of such graphics processing operations, including but not limited to texture sampling logic for performing texture operations, as well as tessellation logic and other vertex processing logic. Processing cluster array 1012 can be configured to execute shader programs related to graphics processing, such as, but not limited to, vertex shaders, tessellation shaders, geometry shaders, and pixel shaders. Parallel processing unit 1002 can transfer data from system memory via I / O unit 1004 for processing. During processing, the transferred data may be stored in on-chip memory (e.g., parallel processor memory 1022) during processing and then written back to system memory.

[0071] When the parallel processing unit 1002 is used to perform graphics processing, the scheduler 1010 can be configured to divide the processing workload into tasks of approximately equal size to better distribute graphics processing operations to multiple clusters 1014A-1014N of the processing cluster array 1012. Each part of the processing cluster array 1012 can be configured to perform different types of processing. For example, a first part can be configured to perform vertex shading and topology generation, a second part can be configured to perform tessellation and geometry shading, and a third part can be configured to perform pixel shading or other screen-space operations to produce a rendered image for display. Intermediate data generated by one or more clusters 1014A-1014N can be stored in a buffer to allow intermediate data to be transferred between clusters 1014A-1014N for further processing.

[0072] Processing cluster array 1012 can receive processing tasks to be executed via scheduler 1010, which receives commands defining the processing tasks from front end 1008. Processing tasks may include indexes of data to be processed, such as surface (patch) data, primitive data, vertex data, and / or pixel data, as well as state parameters and commands defining how to process the data (e.g., which program to execute). Scheduler 1010 can be configured to retrieve the index corresponding to the task, or can receive the index from front end 1008. Front end 1008 can be configured to ensure that processing cluster array 1012 is configured to be active before the workload specified by the incoming command buffer (e.g., batch buffer, push buffer, etc.) is initiated.

[0073] Each instance of one or more instances of parallel processing unit 1002 may be coupled to parallel processor memory 1022 to perform any of the operations described above or elsewhere herein. Parallel processor memory 1022 may be accessed via memory crossbar switch 1016, which may receive memory requests from processing cluster array 1012 and I / O unit 1004. Memory crossbar switch 1016 may access parallel processor memory 1022 via memory interface 1018. Memory interface 1018 may include multiple partition units (e.g., partition units 1020A, 1020B through 1020N), each partition unit may be coupled to a portion (e.g., a memory cell) of parallel processor memory 1022. The number of partition units 1020A-1020N can be configured to be equal to the number of memory units, such that the first partition unit 1020A has a corresponding first memory unit 1024A, the second partition unit 1020B has a corresponding memory unit 1024B, and the Nth partition unit 1020N has a corresponding Nth memory unit 1024N. The number of partition units 1020A-1020N may not be equal to the number of memory units.

[0074] Memory cells 1024A-1024N may include various types of memory devices, including dynamic random access memory (DRAM) or graphics random access memory, such as, but not limited to, synchronous graphics random access memory (SGRAM), which includes graphics double data rate (GDDR) memory. Memory cells 1024A-1024N may also include 3D stacked memory, including but not limited to high-bandwidth memory (HBM), HBM2e, or HDM3. Render targets (e.g., but not limited to framebuffers or texture maps) may be stored in memory cells 1024A-1024N, allowing partitioning cells 1020A-1020N to write portions of each render target in parallel to efficiently utilize the available bandwidth of the parallel processor memory 1022. A local instance of the parallel processor memory 1022 may not be included to support a unified memory design that combines system memory with local cache memory.

[0075] Any cluster 1014A-1014N in the processing cluster array 1012 can process data to be written to any memory cell 1024A-1024N within the parallel processor memory 1022. The memory crossbar switch 1016 can be configured to transfer the output of each cluster 1014A-1014N to any partition cell 1020A-1020N, ​​or to another cluster 1014A-1014N on which additional processing operations can be performed. Each cluster 1014A-1014N can communicate with the memory interface 1018 via the memory crossbar switch 1016 to read from or write to various external memory devices. The memory crossbar switch 1016 can be connected to the memory interface 1018 to communicate with the I / O unit 1004, or to a local instance of the parallel processor memory 1022, enabling processing units within different processing clusters 1014A-1014N to communicate with system memory or other memory local to the non-parallel processing unit 1002. The memory crossbar switch 1016 can use virtual channels to separate traffic flows between clusters 1014A-1014N and partition units 1020A-1020N.

[0076] Multiple instances of the parallel processing unit 1002 can be mounted on a single add-in card, or multiple add-in cards can be interconnected. Even if different instances of the parallel processing unit 1002 have different numbers of processing cores, different amounts of local parallel processor memory, and / or other configuration differences, these different instances can be configured to interoperate. For example, some instances of the parallel processing unit 1002 may include higher-precision floating-point units relative to other instances. Systems including one or more instances of the parallel processing unit 1002 or the parallel processor 1000 can be implemented in a variety of configurations and form factors, including but not limited to desktop, laptop, or handheld personal computers, servers, workstations, game consoles, and / or embedded systems.

[0077] Figure 10A It also includes a block diagram of a partitioning unit 1020 according to at least one embodiment. The partitioning unit 1020 is... Figure 10AAn example of one of the partition units 1020A-1020N in the parallel processor memory. Partition unit 1020 may include L2 cache 1021, frame buffer interface 1025, and ROP 1026 (raster operation unit). L2 cache 1021 may be a read / write cache configured to perform load and store operations received from memory crossbar switch 1016 and ROP 1026. Read misses and urgent write-back requests may be output from L2 cache 1021 to frame buffer interface 1025 for processing. Updates may also be sent to the frame buffer via frame buffer interface 1025 for processing. Frame buffer interface 1025 may interface with one of the memory cells in the parallel processor memory, such as, but not limited to, Figure 10A The memory cells 1024A-1024N (shown as 1024) are located in the parallel processor memory 1022.

[0078] ROP 1026 can be a processing unit that performs raster operations, such as, but not limited to, stenciling, z-testing, blending, etc. ROP 1026 can then output processed graphics data stored in graphics memory. ROP 1026 may include compression logic for compressing depth or color data written to memory and decompressing depth or color data read from memory. The compression logic can be lossless compression logic that utilizes one or more compression algorithms. The type of compression performed by ROP 1026 can vary based on the statistical characteristics of the data to be compressed. For example, incremental color compression is performed on depth and color data on a per-tile basis.

[0079] ROP 1026 can be included in each processing cluster (e.g., Figure 10A The data is stored within clusters 1014A-1014N, not within partition units 1020. Read and write requests for pixel data (not pixel fragment data) can be transferred via memory crossbar switch 1016. Processed graphics data can be displayed on a monitor and routed for further processing by the processor, or routed to... Figure 10A One of the processing entities within the 1000 parallel processors in the system is further processed.

[0080] In at least one embodiment, the parallel processor 1000 may include one or more circuits for executing an API to cause one or more storage locations of one or more scheduled instructions to be indicated to one or more users; executing an API to prevent one or more storage locations of one or more scheduled instructions from being stored; and / or executing regarding Figure 1-7One or more of the processes illustrated or described, or otherwise performing any of the operations described above or elsewhere herein. One or more circuits may be software-configured to execute APIs to cause one or more storage locations of one or more scheduled instructions to be indicated to one or more users; to execute APIs to prevent one or more storage locations of one or more scheduled instructions from being stored; and / or to perform operations related to... Figure 1-7 One or more of the processes illustrated or described, or any of the operations performed above or elsewhere in this document.

[0081] Figure 10B A block diagram including a processing cluster 1014 within a parallel processing unit according to at least one embodiment. The processing cluster may be... Figure 10A An instance of one of the processing clusters 1014A-1014N is provided, which can be used to perform any of the operations described above or elsewhere herein. Processing cluster 1014 can be configured to execute many threads in parallel, where a “thread” refers to an instance of a specific program executed on a specific input dataset. Single Instruction Multiple Data (SIMD) instruction issuing techniques can be used to support the parallel execution of a large number of threads without providing multiple independent instruction units. Single Instruction Multiple Thread (SIMT) techniques can be used to support the parallel execution of a large number of typically synchronous threads using a common instruction unit configured to issue instructions to a set of processing engines within each processing cluster.

[0082] The operation of cluster 1014 can be controlled via pipeline manager 1032, which distributes processing tasks to SIMT parallel processors. Pipeline manager 1032 can... Figure 10A The scheduler 1010 receives instructions and manages the execution of these instructions via the graphics multiprocessor 1034 and / or texture unit 1036. The graphics multiprocessor 1034 may be an example instance of a SIMT parallel processor. However, the processing cluster 1014 may include various types of SIMT parallel processors with different architectures. The processing cluster 1014 may include one or more instances of the graphics multiprocessor 1034. The graphics multiprocessor 1034 can process data and can use the data cross switch 1040 to distribute the processed data to one of several possible destinations, including other shader units. The pipeline manager 1032 can facilitate the distribution of processed data by specifying the destination of the processed data to be distributed via the data cross switch 1040.

[0083] Each graphics multiprocessor 1034 within the processing cluster 1014 may include a set of identical functional execution logic (e.g., arithmetic logic units, load-memory units, etc.) for performing computations for any of the operations described above or elsewhere herein. The functional execution logic can be configured in a pipelined manner, where new instructions can be issued before previous instructions complete. The functional execution logic can support a wide range of operations, including integer and floating-point arithmetic, comparison operations, Boolean operations, bit shifting, and computation of various algebraic functions. Different operations can be performed using the same functional unit hardware, and arbitrary combinations of functional units are possible.

[0084] Instructions transmitted to the processing cluster 1014 can form threads, which may also be called thread bundles, subgroups, waves, or wavefronts. A group of threads executing across a set of parallel processing engines can be called a thread group. Thread groups can execute a common program on different input data. Each thread within a thread group can be assigned to a different processing engine within the graphics multiprocessor 1034. The number of threads in a thread group can be less than the number of processing engines within the graphics multiprocessor 1034. When the number of threads in a thread group is less than the number of processing engines, one or more processing engines may be idle during the processing cycle of that thread group. The number of threads in a thread group can also be more than the number of processing engines within the graphics multiprocessor 1034. When the number of threads in a thread group is more than the number of processing engines within the graphics multiprocessor 1034, processing can be performed in consecutive clock cycles. Multiple thread groups can execute concurrently on the graphics multiprocessor 1034.

[0085] The graphics multiprocessor 1034 includes an internal cache memory for performing load and store operations, such as, but not limited to, any of the operations described above or elsewhere herein. The graphics multiprocessor 1034 may forgo the internal cache and instead use a cache memory within the processing cluster 1014 (e.g., L1 cache 1048). Each graphics multiprocessor 1034 may also access partition units that can be shared across all processing clusters 1014 (e.g., ...). Figure 10A The graphics multiprocessor 1034 has an L2 cache within partition units 1020A-1020N and can be used to transfer data between threads. The graphics multiprocessor 1034 can also access off-chip global memory, which may include one or more of the local parallel processor memory and / or system memory. Any memory outside the parallel processing unit 1002 can be used as global memory. The processing cluster 1014 may include multiple instances of the graphics multiprocessor 1034 and can share common instructions and data, which can be stored in the L1 cache 1048.

[0086] Each processing cluster 1014 may include an MMU 1045 (Memory Management Unit), which can be configured to map virtual addresses to physical addresses. One or more instances of the MMU 1045 may reside in... Figure 10A The MMU 1045 is located within the memory interface 1018. It may include a set of page table entries (PTEs) for mapping virtual addresses to physical addresses on tiles, and optional cache line indexes. The MMU 1045 may include address translation lookup buffers (TLBs) or caches that may reside within the graphics multiprocessor 1034 or L1 1048 cache, or within the processing cluster 1014. Physical addresses can be processed to distribute surface data access locally, allowing for efficient request interleaving between partition units. The cache line indexes can be used to determine whether a request for a cache line is a hit or a miss.

[0087] Processing cluster 1014 can be configured such that each graphics multiprocessor 1034 is coupled to a texture unit 1036 for performing texture mapping operations, such as determining texture sample locations, reading texture data, and filtering texture data. Texture data can be read from an internal texture L1 cache (not shown) or an L1 cache within the graphics multiprocessor 1034, and can be retrieved as needed from an L2 cache, local parallel processor memory, or system memory. Each graphics multiprocessor 1034 can output a processed task to a data crossbar switch 1040 to provide the processed task to another processing cluster 1014 for further processing, or store the processed task in an L2 cache, local parallel processor memory, or system memory via a memory crossbar switch 1016. Pre-ROP 1042 (pre-raster operation unit) can be configured to receive data from the graphics multiprocessor 1034 and direct the data to ROP units, which can be associated with partitioning units (e.g., ...) described herein. Figure 10A The PreROP 1042 unit is located together with partition units 1020A-1020N. The PreROP 1042 unit can perform color blending optimization, organize pixel color data, and perform address translation.

[0088] In at least one embodiment, the processing cluster 1014 may include one or more circuits for executing APIs to cause one or more storage locations of one or more scheduled instructions to be indicated to one or more users; executing APIs to prevent one or more storage locations of one or more scheduled instructions from being stored; and / or executing statements regarding... Figure 1-7One or more of the processes illustrated or described, or otherwise performing any of the operations described above or elsewhere herein. One or more circuits may be software-configured to execute APIs to cause one or more storage locations of one or more scheduled instructions to be indicated to one or more users; to execute APIs to prevent one or more storage locations of one or more scheduled instructions from being stored; and / or to perform operations related to... Figure 1-7 One or more of the processes illustrated or described, or any of the operations performed above or elsewhere in this document.

[0089] Figure 10C A graphics multiprocessor 1034 according to at least one embodiment is illustrated, for example, to perform any of the operations described above or elsewhere herein. The graphics multiprocessor 1034 may be coupled to a pipeline manager 1032 of a processing cluster 1014. The graphics multiprocessor 1034 may include an execution pipeline including, but not limited to, an instruction cache 1052 (e.g., which may store instructions, such as, but not limited to, compiled API instructions), instruction units 1054, address mapping units 1056, register files 1058, one or more general-purpose graphics processing unit (GPGPU) cores 1062, and one or more load / store units 1066, one or more of which may perform load / store operations to load / store instructions corresponding to the execution operations. The GPGPU cores 1062 and load / store units 1066 may be coupled to cache memory 1072 and shared memory 1070 via a memory and cache interconnect 1068. The GPGPU cores 1062 may be part of a SoC, such as, but not limited to, [other components]. Figure 9 It is part of the integrated circuit 900.

[0090] Instruction cache 1052 can receive a stream of instructions to be executed (e.g., perform any of the operations described above or elsewhere herein) from pipeline manager 1032. Instructions can be cached in instruction cache 1052 and dispatched for execution by instruction unit 1054. Instruction unit 1054 can dispatch instructions into thread groups (e.g., thread bundles, subgroups, wavefronts, or waves), with each thread in the thread group assigned to a different execution unit within GPGPU core 1062. Instructions can access any of the local, shared, or global address spaces by specifying an address within a unified address space. Address mapping unit 1056 can be used to translate addresses in the unified address space into different memory addresses accessible by load / store unit 1066.

[0091] Register file 1058 provides a set of registers for the functional units of graphics multiprocessor 1034. Register file 1058 provides temporary storage for operands on data paths connected to functional units of graphics multiprocessor 1034 (e.g., GPGPU core 1062, load / store unit 1066). Register file 1058 can be partitioned among functional units such that each functional unit is allocated a dedicated portion of register file 1058. Register file 1058 can be partitioned among different thread bundles (which may be referred to as wavefronts, subgroups, and / or waves or threads) executed by graphics multiprocessor 1034.

[0092] Each GPGPU core 1062 may include a floating-point unit (FPU) and / or an integer arithmetic logic unit (ALU) for executing instructions of the graphics multiprocessor 1034. The architectures of the GPGPU cores 1062 may be similar or different. A first part of the GPGPU core 1062 may include a single-precision FPU and an integer ALU, while a second part of the GPGPU core may include a double-precision FPU. The FPU may implement IEEE 754-2008 standard floating-point arithmetic or enable variable-precision floating-point arithmetic. The graphics multiprocessor 1034 may also include one or more fixed-function or special-function units for performing specific functions, such as, but not limited to, copying rectangles or pixel blending operations. One or more of the GPGPU cores 1062 may also include fixed-function or special-function logic.

[0093] The GPGPU core 1062 may include SIMD logic capable of executing a single instruction on multiple sets of data. The GPGPU core 1062 can physically execute SIMD4, SIMD8, and SIMD16 instructions, and logically execute SIMD1, SIMD2, and SIMD32 instructions. The SIMD instructions for the GPGPU core may be generated by the shader compiler at compile time, or automatically generated when executing programs written and compiled for Single Program Multiple Data (SPMD) or SIMT architectures. Multiple threads of a program can be configured for a SIMT execution model that can be executed via a single SIMD instruction. For example, eight SIMT threads performing the same or similar operations can be executed in parallel via a single SIMD8 logic unit.

[0094] The memory and cache interconnect 1068 may include an interconnect network that connects each functional unit of the graphics multiprocessor 1034 to the register file 1058 and shared memory 1070. The memory and cache interconnect 1068 may be a cross-switch interconnect that allows the load / store unit 1066 to perform load and store operations between the shared memory 1070 and the register file 1058. The register file 1058 may operate at the same frequency as the GPGPU core 1062, thus data transfer between the GPGPU core 1062 and the register file 1058 can have very low latency. The shared memory 1070 can be used to implement communication between threads executing on functional units within the graphics multiprocessor 1034. The cache memory 1072 can be used as a data cache, for example, for caching texture data transferred between functional units and texture units 1036. The shared memory 1070 can also be used as a program-managed cache. In addition to automatically caching the data stored in the cache memory 1072, threads executing on the GPGPU core 1062 can also programmatically store data in shared memory.

[0095] The parallel processor or GPGPU described herein can be communicatively coupled to a host / processor core to accelerate graphics operations, machine learning operations, pattern analysis operations, and various general-purpose GPU (GPGPU) functions. The GPU can be communicatively coupled to the host processor / core via a bus or other interconnect (e.g., high-speed interconnects, such as, but not limited to, PCIe or NVLink). A System-on-a-Chip (SoC) may include the parallel processor or GPGPU described herein, which executes on the SoC. The GPU may be integrated as a core on a package or chip and communicatively coupled to the core via an internal processor bus / interconnect within the package or chip. Regardless of the GPU's connection method, the processor core can assign work to the GPU in the form of a sequence of commands / instructions contained in a job descriptor. The GPU can then use dedicated circuitry / logic to efficiently process these commands / instructions to perform any of the operations described above or elsewhere herein.

[0096] In at least one embodiment, the graphics multiprocessor 1034 may include one or more circuits for executing an API to cause one or more storage locations of one or more scheduled instructions to be indicated to one or more users; executing an API to prevent one or more storage locations of one or more scheduled instructions from being stored; and / or executing statements regarding... Figure 1-7One or more of the processes illustrated or described, or otherwise performing any of the operations described above or elsewhere herein. One or more circuits may be software-configured to execute APIs to cause one or more storage locations of one or more scheduled instructions to be indicated to one or more users; to execute APIs to prevent one or more storage locations of one or more scheduled instructions from being stored; and / or to perform operations related to... Figure 1-7 One or more of the processes illustrated or described, or any of the operations performed above or elsewhere in this document.

[0097] Figure 11 A processor 1100 according to at least one embodiment is illustrated. The processor 1100 may include a hybrid architecture processor (e.g., Lunar Lake or Meteor Lake) from Intel Corporation, Santa Clara, California, or other processors sharing at least some of the components described herein. The processor 1100 may include one or more central processing units (CPU 1102), one or more graphics processing units (GPU 1106), and / or one or more neural processing units (NPU 1108), which may be, for example, dedicated AI accelerators for offloading artificial intelligence (AI) workloads from the CPU 1102 and GPU 1106. The processor 1100 may use instructions that, if executed, cause the processor 1100 and / or any of its components to perform some or all of the processes and techniques described elsewhere herein. The processor 1100 may include any number of memory and cache units 1110 for facilitating processing between different components of the processor 1100. The memory and cache 1110 on processor 1100 may include one or more levels of cache (e.g., L1, L2, L3, and / or last-level cache) and high-bandwidth memory (e.g., HBM2e or HBM3) in any combination. Regarding processor 1100 and any components described above or elsewhere herein, one or more APIs described herein may, for example, be compiled into instructions that may be fetched by instruction fetching logic or equivalents, decoded by processor decoder or equivalents, scheduled (e.g., sequentially or out of order) for execution by scheduler or equivalents, executed by execution logic or equivalents, reordered, and then retired by retirement logic or equivalents. APIs (and / or compiled instructions including APIs) may be stored in any storage device (e.g., cache and / or memory) internal or external to processor 1100. The results of APIs may be stored in storage devices internal or external to processor 1100, including registers, DRAM, flash memory, SRAM, cache, or other memory. One or more APIs described herein may include calls.

[0098] Processor 1100 may include a computing engine as CPU 1102 and may include any number of cores, such as, but not limited to, up to 16 cores / 22 threads. The cores in CPU 1102 may include P-cores (performance), E-cores (high efficiency), and LP-E cores (low-power, high-efficiency). Performance cores can be used for low-latency, single-threaded, computationally intensive workloads, while high-efficiency cores can be used for multi-threaded, less computationally intensive workloads. Low-power, high-efficiency cores can be used for scalable multi-threaded execution and offloading background tasks. P-cores can be used for single-threaded and limited-threaded execution, while E-cores and LP-E cores are used for multi-threaded throughput and power efficiency.

[0099] The GPU 1106 can include any number of graphics engines, such as, but not limited to, the Intel® Arc™ Graphics Engine (Xe LPG) with 8 Xe cores (up to 128 execution units or EUs). Figure 11 As shown, GPU 1106 may include vector engine 1110 and matrix engine 1112, which, for example, can run FP, INT, and matrix operation tasks simultaneously, individually, or in batches. GPU 1106 may include load / store unit 1114, as well as other memories, such as, but not limited to, instruction cache (I$) 1116 and L1 cache / subsystem local memory (SLM) 1118, which may, for example, store instructions for performing any of the operations described above or elsewhere herein.

[0100] The NPU 1104 may include one or more Intel® AI Boost built-in Neural Processing Units (NPUs). The NPU 1104 can be enumerated as an integrated PCIe device to the host processor. The NPU 1104 may include one or more (e.g., two) Neural Computation Engine (NCE) tiles 1130. Each tile may be configured with any combination of, but not limited to, the following: (e.g., 2000) Multiply-Accumulate (MAC) engines 1134, a post-processing engine (not shown), an AI DSP processor (not shown), and memory per tile (2MB dedicated SRAM), such as... Figure 11As shown. For general computing needs, the neural computing engine 1130 may include a disturbance pipeline 1132, an activation function (AF) 1136, a data transformation 1138, a load / store 1140, and a streaming hybrid architecture vector engine (SHAVE) 1128 for high-performance parallel computing, which may include a DMA (Direct Memory Access) engine 1124 for transporting data between system memory DRAM (Dynamic Random Access Memory) 1126 and a software-managed cache. The built-in device MMU (Memory Management Unit) 1122 plus the IOMMU (Input-Output Memory Management Unit) (not shown) can support multiple concurrent hardware contexts and provide secure isolation between execution contexts according to the MCDM (Microsoft Compute Driver Model) architecture. The processor 1100 may also include a media unit (not shown), which may be included on or separate from the XCD or other components of the processor 1100 to enable video playback and video processing of compressed or uncompressed data, such as hardware-accelerated decoding support for HEVC, AV1, VP9, ​​and AVC, and hardware-accelerated encoding support for HEVC, VP9, ​​and AVC.

[0101] The Intel® Thread Director (which includes firmware built into the processor 1100) prioritizes and manages workload distribution, sending tasks to optimized cores. For example, the Thread Director can tie P-cores, E-cores, and / or LP-E-cores (as described above) together with task scheduling capabilities and the ability to send less demanding tasks to E-cores or LP-E-cores. Intel® Deep Learning Boost (Intel® DL Boost, not shown) provides built-in AI acceleration for training and inference workloads and may include VNNI (for CPU) and DP4a (for GPU) instruction set support. This instruction set can be optimized using the OpenVINO™ toolkit and oneAPI to accelerate INT8 inference. For example, a software stack as described elsewhere in this document can be used to implement AI inference using the OpenVINO™ toolkit. The processor 1100 can be configured to execute applications, such as, but not limited to, CUDA programs.

[0102] In at least one embodiment, the processor 1100 may include one or more circuits for executing an API to cause one or more storage locations of one or more scheduled instructions to be indicated to one or more users; executing an API to prevent one or more storage locations of one or more scheduled instructions from being stored; and / or executing statements regarding... Figure 1-7One or more of the processes illustrated or described, or otherwise performing any of the operations described above or elsewhere herein. One or more circuits may be software-configured to execute APIs to cause one or more storage locations of one or more scheduled instructions to be indicated to one or more users; to execute APIs to prevent one or more storage locations of one or more scheduled instructions from being stored; and / or to perform operations related to... Figure 1-7 One or more of the processes illustrated or described, or any of the operations performed above or elsewhere in this document.

[0103] Processor 1100 may alternatively include a processor based on Qualcomm's AIEngine Direct architecture from Santa Clara, California, or other processors sharing at least some of the components described herein. It may include any number of NPUs, GPUs, CPUs, and other associated components, such as, but not limited to, an NPU 1104 as a Hexagon NPU, a GPU 1106 as an Adreno GPU, a CPU 1102 as a Kryo or Qualcomm Oryon CPU, and a Qualcomm Sensing Hub (not shown) and a memory subsystem 1110. Hexagon NPU 1104 may include power rails, micro-tile inference units, hardware acceleration units, tensor units, scalar units, and vector units (all not shown), which may have dedicated or shared memory (e.g., cache or memory, such as HBM3) for storing, for example, instructions for performing any of the operations described above or elsewhere herein. The Adreno GPU 1106 can provide graphics and parallel processing for AI, in formats including but not limited to 32-bit floating-point (FP32), 16-bit floating-point (FP16), and 8-bit integer (INT8). The Kryo or Qualcomm Oryon CPU 1102 can execute AI workloads and handle the contextualization of ubiquitous generative AI applications. The CPU 1102 may also include an instruction fetch unit, a renaming and dismiss unit, a memory management unit, a vector execution unit, an integer execution unit, and a load and store unit for processing and instruction management. Regarding the processor 1100 and any of its components described above or elsewhere herein, one or more APIs described herein may, for example, be compiled into instructions that may be fetched by the instruction fetch unit, decoded by the processor decoder or equivalent, scheduled (e.g., sequentially or out of order) for execution by the scheduler or equivalent, executed by execution logic or equivalent, reordered, and then dismissed by the renaming and dismiss unit. The API (and / or compiled instructions including the API) can be stored in any storage device (e.g., cache and / or memory) inside or outside the processor 1100. An arbitrary number of CPU cores 1102 can be included in an arbitrary number of CPU clusters, which can be coupled to memory and / or cache, such as, but not limited to, a shared L2 cache. Memory can be separate or shared; for example, the CPU clusters of CPU cores 1102 can be coupled to a memory subsystem 1110, which can include structures capable of reading and writing to memory (e.g., DRAM), system-level caches, and an arbitrary number of memory management units.The Qualcomm sensing hub (not shown) includes a miniature NPU, power rails, and conventional sensors (such as gyroscopes, accelerometers, or even barometers) that support voice and data streaming. The memory subsystem 1110 may include memory and cache on the processor 1100, which may include L1 or more levels of cache (e.g., L1, L2, L3, and / or last-level cache) and high-bandwidth memory (e.g., HBM2e or HBM3) in any combination, for example, for storing information and / or instructions for performing any of the operations described above or elsewhere herein. All or part of the memory and / or cache in the memory subsystem 1110 may be shared or used individually by any component or combination of components on the processor 1100 (e.g., GPU 1106, NPU 1104, and CPU 1102).

[0104] The Qualcomm AI Engine 1100 can be programmed and controlled using a software stack to perform some or all of the operations described herein, including, for example, versions of the Qualcomm® Neural Processing SDK for inference on Android, Linux, and Windows. Developer libraries and services support programming languages, virtual platforms, and compilers. At lower levels of the software stack, system software includes a basic real-time operating system (RTOS), system interfaces, and drivers. The software stack supports various operating systems, including Android, Windows, Linux, and QNX, as well as deployment and monitoring infrastructures such as Prometheus, Kubernetes, and Docker. For direct cross-platform access to the GPU 1106, OpenCL and DirectML are supported. For the CPU 1102, LLVM compiler infrastructure optimizations enable accelerated and efficient AI inference. Regarding the Qualcomm AI Engine 1100 and any components described above or elsewhere herein, one or more APIs described herein may, for example, be compiled into instructions that may be fetched by instruction fetching logic or equivalents, decoded by processor decoders or equivalents, scheduled (e.g., sequentially or out of order) by a scheduler or equivalent for execution, executed by execution logic or equivalents, reordered, and then retired by retirement logic or equivalents. The APIs (and / or compiled instructions including the APIs) may be stored in any storage device (e.g., cache and / or memory) internal or external to the Qualcomm AI Engine 1100. The results of the APIs may be stored in storage devices internal or external to the Qualcomm AI Engine 1100, including registers, DRAM, flash memory, SRAM, cache, or other memory.

[0105] In at least one embodiment, the processor 1100 or the Qualcomm AI engine 1100 may include one or more circuits for executing an API to cause one or more storage locations of one or more scheduled instructions to be indicated to one or more users; executing an API to prevent one or more storage locations of one or more scheduled instructions from being stored; and / or executing statements regarding... Figure 1-7 One or more of the processes illustrated or described, or otherwise performing any of the operations described above or elsewhere herein. One or more circuits may be software-configured to execute APIs to cause one or more storage locations of one or more scheduled instructions to be indicated to one or more users; to execute APIs to prevent one or more storage locations of one or more scheduled instructions from being stored; and / or to perform operations related to... Figure 1-7 One or more of the processes illustrated or described, or any of the operations performed above or elsewhere in this document.

[0106] Figure 12AA processor 1200 according to at least one embodiment is illustrated. The processor 1200 may include a Scalable family processor from Intel Corporation, Santa Clara, California, or other processors that share at least some of the components described herein. The processor 1200 may include one or more cores 1212(1)-1212(N) capable of performing the operations described elsewhere herein, where N is any integer greater than 1. Cores 1212(1)-1212(N) may be interconnected using ring and / or mesh interconnects. Utilizing a mesh interconnect architecture, arrays of vertical and horizontal communication paths may allow traversal from one core to another 1212(1)-1212(N) via the shortest path (jumping to the correct row along the vertical path and to the correct column along the horizontal path). For the mesh interconnect, a die may accommodate cores 1212(1)-1212(N) and may include a Converged Mesh Stop Point (CMS) grid that may be associated with cores 1212(1)-1212(N) (e.g., 1:1). Each core may be associated with a low-level cache (LLC) slice 1214(1)-1214(N), or cores 1212(1)-1212(N) may share a cache, such as a low-level cache. LLC 1214(1)-1214(N) may be inclusive or non-inclusive (having blocks not present in the high-level cache) by merging blocks from a higher-level cache (e.g., L2 cache). Each core and LLC slice may include a caching and home agent (CHA) (not shown) that maintains cache coherency by providing resource scalability via mesh interconnect, thereby enabling cache coherency features of Intel® HyperPath Interconnect (Intel® UPI 1216). UPI 1216 provides a coherent interconnect for scalable systems and allows multiple processors to share a single shared address space via links, such as, but not limited to, two or three UPI links per processor.

[0107] Processor 1200 may also include system agent 1210, which may house and / or perform various functions, such as, but not limited to, memory management, display functions, and / or input / output (I / O) functions. For example, processor 1200 may include one or more integrated memory controllers (IMCs) 1208. IMC 1208 may control and manage memory, such as, but not limited to, different memory types, such as DDR RAM, such as DDR4, or other memory described elsewhere herein. System agent 1210 may include a display controller (not shown) for supporting one or more displays. System agent 1210 may also integrate a PCIe 1204 (e.g., up to 20 PCIe lanes), which may, for example, be connected to an external dedicated graphics connector via a DMI bus (e.g., Intel's DMI 3.0 bus) 1206. System agent 1210 may include an image processing unit (IPU) (not shown) that integrates an on-die image signal processor (ISP). Structure 1202 provides scalability for connecting to other nodes (e.g., processors, such as processor 1200) and can be used, for example, with Cornelis Networks (an element of the Intel® Scalable System Framework), which provides performance for high-performance computing (HPC) workloads and the ability to scale to tens of thousands of nodes.

[0108] Figure 12B Components within a core 1212 according to at least one embodiment are illustrated. The core 1212 may include a front-end 1218, a back-end or execution engine 1232, and a memory subsystem 1242. The front-end 1218 may provide operations (e.g., operations described elsewhere herein) to the execution engine 1232 by decoding instructions stored in memory. For example, the front-end 1218 may include micro-operation (µOps) cache paths and / or traditional paths, and a branch prediction unit 1221 capable of determining path instructions. A traditional path of instructions may include fetching variable-length (e.g., x86) instructions from an L1 instruction cache 1220 and instruction fetch and pre-decode 1222, queuing these instructions into an instruction queue 1224, and decoding the instructions into µOps that can be provided to an allocation queue 1228 using a decoder 1226. Alternatively, the µOps cache path may include a cache that includes decoded µOps (µOps 1230) that can be sent to the allocation queue 1228. The allocation queue 1228 can act as an interface between the front end 1218 and the execution engine 1232, and can provide instructions to the execution engine 1232. For example, one or more APIs described herein can be compiled into instructions that can be stored, processed, and executed by the front end 1218 and the execution engine 1232, and stored in the memory subsystem 1242.

[0109] Execution engine 1232 can receive micro-operations into reordering buffer 1234, which can register allocations, renames, and decommissions of µOPs. µOPs can be sent from the reordering buffer to scheduler 1236, which can be connected to one or more different execution units 1238, which can be connected to address generation units (AGUs) 1240. Execution units 1238 can perform operations such as basic arithmetic logic unit (ALU) operations, multiplication, division, and / or more complex operations, such as, but not limited to, various vector operations. Scheduler 1236 can manage the queuing of µOPs for one or more execution units 1238 based on, for example, the operations that need to be performed.

[0110] The memory subsystem 1242 can handle load and store requests as well as sorting operations. For example, µOPs may be associated with memory accesses (e.g., load and store), and these µOPs can be sent through dedicated scheduler ports that can perform these memory operations. For example, store and load operations can be sent to load and store buffers 1244. The memory subsystem 1242 may also include shared or separate L1 data and instruction caches 1246, and an L2 cache 1248 that can be used and shared by the L1 data and instruction caches 1246. (As described above regarding...) Figure 12A Each core 1212 can be connected to a slice of a third-level cache (e.g., LLC1214), which can be shared by all cores 1212.

[0111] In at least one embodiment, the processor 1200 may include one or more circuits for executing an API to cause one or more storage locations of one or more scheduled instructions to be indicated to one or more users; executing an API to prevent one or more storage locations of one or more scheduled instructions from being stored; and / or executing statements regarding... Figure 1-7 One or more of the processes illustrated or described, or otherwise performing any of the operations described above or elsewhere herein. One or more circuits may be software-configured to execute APIs to cause one or more storage locations of one or more scheduled instructions to be indicated to one or more users; to execute APIs to prevent one or more storage locations of one or more scheduled instructions from being stored; and / or to perform operations related to... Figure 1-7 One or more of the processes illustrated or described, or any of the operations performed above or elsewhere in this document.

[0112] Figure 13An AI accelerator 1300 according to at least one embodiment is illustrated. Processor 1300 may include a processor with an AI accelerator architecture manufactured by Intel Corporation, Santa Clara, California, or other processors sharing at least some of the components described herein. AI accelerator 1300 may use instructions that, if executed by AI accelerator 1300, cause AI accelerator 1300 to perform some or all of the processes and techniques described elsewhere herein. For example, with respect to AI accelerator 1300 and any components described above or elsewhere herein, one or more APIs described herein may, for example, be compiled into instructions that may be fetched by instruction fetching logic or equivalents, decoded by processor decoders or equivalents, scheduled (e.g., sequentially or out of order) by a scheduler or equivalent for execution, executed by execution logic or equivalents, reordered, and then retired by retirement logic or equivalents. APIs (and / or compiled instructions including APIs) may be stored in any storage device internal or external to AI accelerator 1300 (e.g., in cache and / or memory). The results of the API can be stored in internal or external storage devices of the AI ​​accelerator 1300, including registers, DRAM, flash memory, SRAM, cache, or other memory. The AI ​​accelerator 1300 may include one or more compute dies, which may include homogeneous or heterogeneous processors. The compute dies may include one or more central processing units (CPUs), one or more graphics processing units (GPUs), or a combination of both.

[0113] In at least one embodiment, the computational die may include a computational engine for performing AI computations. In at least one embodiment, the computational die of the AI ​​accelerator 1300 may be split into any number (e.g., four) clusters, which may be referred to as DCORE (Deep Learning Core) 1306, and include any number of matrix multiplication engines (MME) 1308, tensor processor cores (TPC) 1310, memory management units 1312, and L2 caches 1314 in any combination. The MME 1308 may perform operations using matrix multiplication, such as fully connected layers, convolutions, and batch general matrix multiplication (GEMM). The MME 1308 may be equipped with a multiplication-accumulation unit (MAC) (not shown), which may perform general matrix multiplication (GEMM) operations, such as, but not limited to, AxB multiplication, which involves generating a tensor C [NxM] from two input tensors A [NxK] and B [KxN]. The MME 1308 may be programmed with array dimensions, positions, data types, and various operands. The MME 1308 can retrieve tensors A and B from memory and pull them into its streaming buffer for parallel matrix multiplication by the MAC. After completion, the MME 1308 can push tensor C back to memory. The TPC 1310 may include any number of scalar units for performing scalar operations, any number of vector units for performing vector operations, any number of register files or local memory units (e.g., vector local memory), and load and store components for instructions, which may be coupled to memory or caches (e.g., HBM, L3 cache, and / or L2 cache) (all not shown). The TPC can support different types of parallel processing, such as Very Long Instruction Word (VLIW) Single Instruction Multiple Data (SIMD) data types such as, but not limited to, FP32, BF16, FP16, and FP8 (both E4M3 and E5M2), UINT32, INT32, UINT16, INT16, UINT8, and INT8 data types. Any number of computational dies can be interconnected. Interconnects that can connect computing dies can be via intermediate bridges, for example, those intermediate bridges that are transparent to software.

[0114] The memory on the AI ​​accelerator 1300 may include one or more levels of cache (e.g., L1, L2, L3, and / or last-level cache) and high-bandwidth memory (e.g., HBM2e or HBM3) in any combination. The memory and / or cache system may be unified or separate. The compute die of the AI ​​accelerator 1300 may include on-chip memory comprising one or more levels of cache (e.g., two levels). On-chip SRAM or other memory described elsewhere herein may be used as a unified last-level cache (L3) or split into multiple slices of L2 cache accessible to the MME 1308 and TPC 1310 group. Using on-chip memory as an L2 or L3 cache can be fully configurable by software, which can dynamically determine its optimal cache allocation based on I / O tensors. The AI ​​accelerator 1300 may include one or more memory management units (MMUs) 1322 for managing memory, such as allowing the AI ​​accelerator 1300 memory subsystem to operate in virtual space when accessing VRAM.

[0115] AI accelerator 1300 may include a communication port (e.g., a PCIe Gen5 x16 port) 1302 for communicating with a host and scheduling and synchronization unit 1304. AI accelerator 1300 may include a media unit 1316, which may include any number or combination of media decoder engines (DECs) 1320 and rotation engines (ROTs) 1318. AI accelerator 1300 may include a network unit 1324, which may include any number or combination of network ports 1326 and an accompanying RDMA engine 1328, L2 cache, and memory (e.g., HBM2e or HBM3) stack. AI accelerator 1300 may include a programmable control path entity (not shown) for managing the parallel and efficient execution of the various engines. The control path may include a submission queue (SQ) that can be issued by the runtime system, a completion queue (CQ) that can be used for job completion reporting, a programmable scheduling mechanism that can be used for task scheduling, a programmable hardware synchronization mechanism or "synchronization manager (SM)" that can be used for hardware synchronization, and a programmable interrupt service mechanism or "interrupt manager (INTR)" that can pass asynchronous events to the driver.

[0116] AI Accelerator 1300 may include media decoding units supporting video formats such as, but not limited to, HEVC, Progressive H.264, SVC base layer, MVC, VP9, ​​JPEG, and Progressive JPEG. AI Accelerator 1300 may support post-processing of the decoded media stream, such as, but not limited to, image downsizing (image resizing), vertical and horizontal scaling at different scaling ratios, image enlargement, image cropping, bilinear scaling, and Lancos scaling. AI Accelerator 1300 may implement two post-processing channels per decoder unit, one for scalar (up and down) and the other solely for outputting the original image. AI Accelerator 1300 may include a hardware rotation engine that performs the following transformations on the input image: 2D rotation, 3D rotation, projection, image warping and de-warping, resampling of the input data at user-defined coordinates, and rescaling.

[0117] The AI ​​accelerator 1300's RDMA 1328 based on converged Ethernet enables scaling from a single node (i.e., from a single AI accelerator 1300 to hundreds or thousands of nodes or AI accelerators 1300). The network subsystem 1324 may include the Intel® Gaudi® Communications Library (IGCL), a master controller coordinating data movement, and a programmable scheduling mechanism that enables smooth engine activation while maintaining task dependencies. The accelerator network subsystem may include a Gigabit Ethernet NIC port 1326, a Layer 2 MAC (not shown), and the RDMA engine 1328. The AI ​​accelerator 1300 may include an aggregation engine for performing summation activities. All engines in the processor 1300 can run in parallel; for example, the MME 1308, TPC 1310, and NIC 1326 can all operate simultaneously. Dependencies may exist between operations running on different engines; for example, the output of one engine may be used as the input of another engine, and / or the MME, TPC, and NIC may be scheduled to run in parallel. Once one engine has finished its operation, another engine can be scheduled to begin working on the next operation (executed immediately once its input is ready).

[0118] The AI ​​accelerator 1300 can be operated and controlled using a software layer 1328, which may include low-level components such as, but not limited to, a graph compiler, an automatic kernel fusionist and pre-compiled kernel libraries, and integrations with the AI ​​ecosystem such as, but not limited to, PyTorch, DeepSpeed, Hugging Face, vLLM, Ray, etc., or as described elsewhere in this document regarding software and programming platforms. The software layer 1328 may include implementations of algorithms such as, but not limited to, PagedAttention, FlashAttention, etc. The software layer 1328 may generate optimized binary code that implements a given model topology, such as, but not limited to, performing operator fusion, data layout management, parallelization, pipeline and memory management, and graph-level optimization.

[0119] In at least one embodiment, the AI ​​accelerator 1300 may include one or more circuits for executing an API to cause one or more storage locations of one or more scheduled instructions to be indicated to one or more users; executing an API to prevent one or more storage locations of one or more scheduled instructions from being stored; and / or executing regarding Figure 1-7 One or more of the processes illustrated or described, or otherwise performing any of the operations described above or elsewhere herein. One or more circuits may be software-configured to execute APIs to cause one or more storage locations of one or more scheduled instructions to be indicated to one or more users; to execute APIs to prevent one or more storage locations of one or more scheduled instructions from being stored; and / or to perform operations related to... Figure 1-7 One or more of the processes illustrated or described, or any of the operations performed above or elsewhere in this document.

[0120] This paper describes a neuromorphic computing system employing a multi-core architecture, where each core houses computing elements including neurons, synapses with on-chip learning capabilities, and local memory for storing synaptic weights and routing tables. Figure 14This is a simplified block diagram 1400 illustrating at least a portion of an example of such a neuromorphic computing device 1405 according to at least one embodiment. The neuromorphic computing device 1405 may include a neuromorphic processor from Intel Corporation, Santa Clara, California, or other processors that include at least a portion of the components described herein. As shown in this example, the device 1405 may be equipped with a network 1410 consisting of multiple neural network cores interconnected by a network on the device, thereby potentially defining multiple distinct connections between the cores. For example, the device 1405 may provide a network 1410 of spiking neural network cores, each core communicating via short packet pulse messages sent from one core to another through network channels. Each core (e.g., 1415) may have processing and memory resources, as well as logic, for implementing a number of primitive nonlinear time computation elements, such as, but not limited to, multiple (e.g., more than 1000) distinct artificial neurons (referred to herein as “neurons”). For example, each core may be able to implement multiple neurons concurrently, allowing the neuromorphic core to implement many, many neurons using the device 1405. With respect to neuromorphic computing device 1405 and any components described above or elsewhere herein, one or more APIs or equivalents described herein may, for example, be compiled into instructions or equivalents that may be fetched by instruction fetching logic or equivalents, decoded by processor decoder or equivalents, scheduled (e.g., sequentially or out of order) by scheduler or equivalents for execution, executed by execution logic or equivalents, reordered, and then retired by retirement logic or equivalents. APIs (and / or compiled instructions including APIs) may be stored in any storage device (e.g., cache and / or memory) internal or external to neuromorphic computing device 1405. The results of the APIs may be stored in storage devices internal or external to neuromorphic computing device 1405, including registers, DRAM, flash memory, SRAM, cache, or other memory equivalents.

[0121] continue Figure 14For example, neuromorphic computing device 1405 may also include processor 1420 and system memory 1425 for implementing one or more components to manage and provide the functionality of neuromorphic computing device 1405. For instance, a system manager 1430 may be provided to manage the global attributes and operations of neuromorphic computing device 1405 (e.g., attributes affecting core network 1410, multiple cores in network 1410, interconnection of neuromorphic computing device 1405 with other devices, managing access to global system memory 1425, and other potential examples). In one example, system manager 1430 may manage the definition and configuration of specific routing tables for individual routers in network 1410, orchestration of network definitions and attributes to be applied to network 1410 (e.g., weights, attenuation rates, etc.), core synchronization and time multiplexing management, routing input to appropriate cores, and other potential functions.

[0122] As another example, the neuromorphic computing device 1405 may also include a programming interface 1435 through which a user or system can specify the neural network definition to be applied (e.g., via routing tables and individual neuron attributes), implemented by the neuromorphic core grid 1410. A software-based programming tool may be provided with or separately from the neuromorphic computing device 1405, through which a user can provide a definition for a specific neural network to be implemented using the neuromorphic core network 1410. The programming interface 1435 can receive input from a programmer, then generate the corresponding routing table and populate the specified parameters into the local memory of each neuromorphic core (e.g., 1415) to implement the corresponding custom artificial neural network implemented by the neuromorphic core 1415.

[0123] In certain circumstances, neuromorphic computing device 1405 can advantageously engage and interoperate with other devices, including general-purpose computing devices, to enable specific applications and use cases. Therefore, external interface logic 1440 may be provided in certain situations to communicate with one or more other devices (e.g., via one or more defined communication protocols). External interface 1440 may be used to accept input data from another device or an external memory controller used as an input data source. External interface 1440 may additionally or alternatively be used to allow the results or outputs of computations performed using the neural network implemented using neuromorphic computing device 1405 to be provided to another device (e.g., another general-purpose processor implementing machine learning algorithms) to enable additional applications and enhancements, among other examples.

[0124] like Figure 14The diagram illustrates a network 1410 interconnected by a network on the device, showing a portion of a network structure interconnecting multiple neuromorphic cores (e.g., 1415a-d). For example, several neuromorphic cores (e.g., 1415a-d) can be provided in a mesh, each core interconnected via a network including multiple routers (e.g., 1450). In one implementation, each neuromorphic core (e.g., 1415a-d) can be connected to a single router (e.g., 1450) in the routers, and the routers can be connected to at least one other router (e.g., [other routers]). Figure 14 (As shown at 1410 in the diagram). As an example, in one particular implementation, four neuromorphic cores (e.g., 1415a-d) can be connected to a single router (e.g., 1450), and each router 1450 can be connected to two or more other routers to form a multi-core mesh, thereby allowing each neuromorphic core to interconnect with every other neuromorphic core in the neuromorphic computing device 1405. Furthermore, since each neuromorphic core can be configured to implement multiple different neurons, the router network of the neuromorphic computing device 1405 can similarly implement connections or artificial synapses (or simply "synapses") defined between any two of the potential many (e.g., 30,000+) neurons defined by the network definition using the neuromorphic cores 1410 provided in the neuromorphic computing device 1405.

[0125] Figure 14A block diagram of the internal components of an example implementation of the neuromorphic core 1415 is shown. In one example, a single neuromorphic core may implement a number of neurons (e.g., 1024), which share the architectural resources of the neuromorphic core 1415 in a time-division multiplexing manner. In one example, each neuromorphic core 1415 may include a processor block 1455 capable of executing arithmetic functions and routing related to the implementation of the digitally implemented artificial neurons, such as, but not limited to, those explained herein. Each neuromorphic core 1415 may also provide local memory in which routing tables of the neural network can be stored and accessed, accumulated potentials of each cell body of each neuron implemented using core 1415 can be tracked, parameters of each neuron implemented by core 1415 can be recorded, and other data and usage can be recorded. Components or architectural resources of the neuromorphic core 1415 may also include: an input interface 1465 for receiving input spike messages generated by other neurons on other neuromorphic cores; and an output interface 1470 for sending spike messages to other neuromorphic cores via a mesh network 1410. In some instances, the routing logic of the neuromorphic core 1415 can be implemented at least partially using the output interface 1470. Furthermore, in some cases, the core (e.g., 1415) can implement multiple neurons within an example SNN, and some of these neurons can be interconnected. In this case, spiking messages sent between neurons hosted on the core 1415 can forgo communication via the routing structure of the neuromorphic computing device 1405 and can be managed locally within the specific neuromorphic core 1415.

[0126] Each neuromorphic core may also include logic for implementing artificial dendrites 1480 and artificial cell bodies 1485 (hereinafter referred to as “dendrites” and “cell bodies”, respectively) for each neuron 1475. Dendrite 1480 may be a hardware-implemented process for receiving impulses from network 1410. Cell body 1485 may be a hardware-implemented process for receiving the current time-accumulated neurotransmitter mass of each dendrite and evolving the potential states of each dendrite and cell body to generate outgoing impulse messages at appropriate times. Dendrite 1480 may be defined for each connection receiving input from another source (e.g., another neuron). In one implementation, the dendritic process 1480 may receive and process the impulse message as it arrives serially from network 1410 in a time-division multiplexed manner. With the reception of impulses, neuronal activation (tracked using cell body 1485 (and local memory 1460)) may increase. When the activation of a neuron 1475 exceeds a threshold set for neuron 1475, neuron 1475 generates a spike message, which is propagated via output interface 1470 to a fixed set of fan-out neurons. The network distributes the spike messages to all destination neurons, which in turn can update their activation in a transient, time-dependent manner in response. This can lead to some of the destination neurons also exceeding their corresponding thresholds and triggering further spike messages, just as in real biological neural networks.

[0127] As described above, the neuromorphic computing device 1405 can reliably implement spiking-based neural computing models. Such models are also referred to as spiking neural networks (SNNs). In addition to neuronal and synaptic states, SNNs incorporate temporal concepts. For example, in SNNs, communication occurs via event-driven action potentials or spiking, which convey no explicit information other than the spiking time and the implicit source and destination neuron pairs corresponding to the spiking transmission. The computation of the result of a dynamic nonlinear integral as a weighted spiking input occurs in each neuron. In some implementations, recurrent and dynamic feedback can be incorporated into the SNN computation model. Furthermore, various network connectivity models can be employed to model a wide range of real-world networks or relationships, including fully connected (all-to-all) networks, feedforward trees, fully random projections, "small-world" networks, and other examples. Isomorphic two-dimensional networks at the neuromorphic core (e.g., but not limited to...) Figure 14The network shown in the example can advantageously support all these network models. Since some or all of the cores of the neuromorphic computing device 1405 can be connected, some or all of the neurons defined in a core can also be fully connected via a certain number of router hops. The neuromorphic computing device 1405 may also include fully configurable routing tables for defining various neural networks by allowing neurons in each core to distribute their spurs to any number of cores in the grid 1410 to achieve a completely arbitrary connection graph.

[0128] In improved implementations of systems capable of supporting SNNs, for example, but not limited to... Figure 14 The example illustrates a very large-scale integrated circuit (VLSI) hardware device that can provide high-speed, reliable circuitry to implement SNNs (Spiritual Neural Networks) to model the information processing algorithms used by the brain, but in a more programmable way. For instance, while a biological brain can only perform a specific set of defined behaviors (a consequence of years of development), a neuromorphic processor device can provide the ability to rapidly reprogram all neural parameters. Therefore, a single neuromorphic processor can be used to implement a wider range of behaviors than a single slice of biological brain tissue. This distinction can be achieved by employing neuromorphic processors with neuromorphic designs that are distinctly different from those found in natural neural circuitry.

[0129] As an example, a neuromorphic processor can implement a spontaneous neural network (SNN) using time-multiplexed computation in both a spiking communication network and the neuronal mechanism of the neuromorphic computing device 1405. Therefore, the physical circuitry of the neuromorphic computing device 1405 can be shared by many neurons to achieve a higher neuron density. Through time multiplexing, the network can connect N cores with a total wiring length of O(N), while the length of discrete point-to-point wiring will be extended to O(N). 2 This significantly reduces wiring resources to accommodate planar and non-plastic VLSI routing techniques, among other examples. In the neuromorphic core, time multiplexing can be implemented through dense memory allocation, for example, using static random access memory (SRAM) with a shared bus, address decoding logic, and other multiplexed logic elements. The state of each neuron can be stored in the processor's memory, where data describing the state of each neuron includes the state of the collective synapse of each neuron, all currents and voltages on its membrane, and other example information (e.g., but not limited to configuration and other information).

[0130] Neuromorphic processors can be implemented in a “digital” manner, unlike other processors that employ more “analog” or “isomorphic” neuromorphic approaches. For example, a digital implementation can use digital adder and multiplier circuitry to integrate synaptic currents, in contrast to an analog isomorphic neuromorphic approach that accumulates charge on capacitors in a manner similar to how neurons accumulate synaptic charge on their lipid membranes. For instance, the accumulated synaptic charge for each neuron can be stored in the local memory of the corresponding core. Furthermore, at the architectural level of an example digital neuromorphic processor, reliable and deterministic operation can be achieved through time synchronization across the core network, ensuring that any two executions of the design, given the same initial conditions and configuration, will produce the same results. Asynchronicity can be reserved at the circuit level to allow individual cores to operate as quickly and freely as possible while maintaining determinism at the system level. Therefore, in neural computing, the concept of time as a time variable can be abstracted away from the “wall clock” time used by the hardware to perform computations. Thus, in some implementations, a time synchronization mechanism can be provided that globally synchronizes the neuromorphic cores at discrete time intervals. The synchronization mechanism allows neural computation to be completed at the fastest speed allowed by the circuit, and there is a difference between the runtime and the biological time for modeling neuromorphic systems.

[0131] In operation, the neuromorphic computing device 1405 can start in an idle state when all neuromorphic cores are inactive. As each core asynchronously loops through its neurons, it generates impulse messages, which are routed by the mesh interconnect to the appropriate destination core containing all destination neurons. The implementation of multiple neurons on a single neuromorphic core can be time-multiplexed, and time steps can be defined, where all impulses involving multiple neurons can be processed and considered using the shared resources of the respective cores. When each core completes its service to its neurons within the corresponding time step, in some implementations, the core can communicate with neighboring cores using synchronization messages (e.g., using a handshake) to refresh the mesh of all transmitted impulse messages, allowing the core to safely determine that all impulses have been serviced within a certain time step. At this point, all cores can be considered synchronized, allowing them to advance their time steps and return to the initial state to begin the next time step.

[0132] Given this context, as described above, a device (e.g., 1405) can be provided to realize an interconnected neuromorphic core grid 1410, wherein core 1415 can realize multiple artificial neurons capable of interconnecting to realize an SNN. Each neuromorphic core (e.g., 1415) can provide two loosely coupled asynchronous processes: an input dendrite process (e.g., 1480) that receives impulses from network 1410 and applies them to the appropriate destination dendritic compartment at an appropriate future time; and an output cell body process (e.g., 1485) that receives the current-time accumulated neurotransmitter mass of each dendritic compartment and evolves the membrane potential state of each dendrite and cell body to generate an outgoing impulse message at an appropriate time (e.g., when the threshold potential of the cell body is reached). It should be noted that, from a biological perspective, the names of dendrites and cell bodies used herein are only approximate to the function of these features and should not be interpreted too literally.

[0133] In at least one embodiment, the neuromorphic computing device 1405 may include one or more circuits for executing an API to cause one or more storage locations of one or more scheduled instructions to be indicated to one or more users; executing an API to prevent one or more storage locations of one or more scheduled instructions from being stored; and / or executing statements regarding... Figure 1-7 One or more of the processes illustrated or described, or otherwise performing any of the operations described above or elsewhere herein. One or more circuits may be software-configured to execute APIs to cause one or more storage locations of one or more scheduled instructions to be indicated to one or more users; to execute APIs to prevent one or more storage locations of one or more scheduled instructions from being stored; and / or to perform operations related to... Figure 1-7 One or more of the processes illustrated or described, or any of the operations performed above or elsewhere in this document.

[0134] Figure 15 This is a block diagram of an embodiment of a multi-node network capable of enabling remote memory computing according to any embodiment. System 1500 may represent the node network described herein, for example, the node network can be used to perform some or all of the operations described herein. System 1500 may represent a data center. System 1500 may represent a server farm. System 1500 may represent a data cloud or processing cloud. System 1500 may represent a supercomputer. System 1500 may include tens, hundreds, or thousands of nodes. The nodes of System 1500 may include processors, such as, but not limited to, a central processing unit (CPU), a graphics processing unit (GPU), or any combination of processors described herein, such as, but not limited to, processors described herein. Figure 9-21BOther processors in the system. For any processor in System 1500 and any components described above or elsewhere herein, one or more APIs or equivalents described herein may, for example, be compiled into instructions or equivalents that may be fetched by instruction fetching logic or equivalents, decoded by processor decoder or equivalents, scheduled (e.g., sequentially or out of order) for execution by scheduler or equivalents, executed by execution logic or equivalents, reordered, and then retired by retirement logic or equivalents. APIs (and / or compiled instructions including APIs) may be stored in any storage device (e.g., cache and / or memory) inside or outside the processor or node. The results of APIs may be stored in storage devices inside or outside the processor or node, including registers, DRAM, flash memory, SRAM, cache, or other memory equivalents. System 1500 may include more than nine thousand nodes, each node comprising two Intel Xeon Max processors, six Intel Max series GPUs, and a unified memory architecture, such as, but not limited to, the architecture used in Intel's Aurora supercomputer in Santa Clara, California, or other supercomputers that share at least some of the components described herein.

[0135] One or more clients 1502 send requests to system 1500 via network 1504. Network 1504 represents one or more local area networks, wide area networks, or a combination of both. Clients 1502 can be human or machine clients that generate requests for operations to be performed by system 1500. System 1500 executes the application or data computation task requested by client 1502.

[0136] System 1500 may include one or more racks, which represent structural and interconnect resources for housing and interconnecting multiple computing nodes. Rack 1510 may include multiple nodes 1530. Rack 1510 may carry multiple blade assemblies 1520(0)-1520(N-1), where N is an integer greater than or equal to 2. Carrying may refer to providing power, structural or mechanical support, and interconnection. Blades 1520(0)-1520(N-1) may refer to computing resources on a printed circuit board (PCB), where the PCB houses the hardware components of one or more nodes 1530. Blades 1520(0)-1520(N-1) may include or exclude a chassis, enclosure, or other “box” besides those provided by rack 1510. Blades 1520(0)-1520(N-1) may include an enclosure with exposed connectors for connection to rack 1510. System 1500 may or may not include rack 1510, and each blade (e.g., 1520(0)) may include a chassis or housing that may be stacked or otherwise closely proximate with other blades and allow nodes 1530 to interconnect. System 1500 may include 10,624 compute blades, comprising 63,744 Intel Max series GPUs and 21,248 Intel Xeon Max CPUs on 166 racks.

[0137] System 1500 may include architecture 1570, which represents one or more interconnectors of nodes 1530. Architecture 1570 may include multiple switches 1572 or routers or other hardware for routing signaling between nodes 1530. Furthermore, architecture 1570 may couple system 1500 to network 1504 for access by client 1502. In addition to routing devices, architecture 1570 may also be considered to include cables or ports or other hardware devices for coupling nodes 1530 together. Architecture 1570 may have one or more associated protocols for managing signaling routing through system 1500. One or more protocols are at least partially dependent on the hardware devices used in system 1500.

[0138] As shown in the figure, rack 1510 may include N blades (e.g., 1520(0)-1520(N-1)). In addition to rack 1510, system 1500 may also include rack 1550. As shown in the figure, rack 1550 may include M blades (e.g., 1560(0)-1560(M-1)). M is not necessarily the same as N; therefore, it is understood that various different hardware device components may be used and coupled together into system 1500 via structure 1570. Blades 1560(0)-1560(M-1) may be the same as or similar to blades 1520(0)-1520(N-1). Node 1530 may be any type of node described herein and is not necessarily of the same type. System 1500 is not limited to homogeneous or non-homogeneous systems.

[0139] The nodes in blade 1520(0) are shown in detail. However, other nodes in system 1500 may be the same or similar. At least some nodes 1530 may be compute nodes, having processor 1532 and memory 1540. A compute node is a node having processing resources (e.g., one or more processors) that executes an operating system and can receive and process one or more tasks. At least some nodes 1530 may include storage server nodes, which have servers as processing resources 1532 and memory 1540. A storage server is a node having more storage resources than a compute node, and instead of having processors for performing tasks, a storage server includes processing resources for managing access to storage nodes within the storage server.

[0140] Node 1530 may include interface controller 1534, which may represent logic for controlling node 1530's access to structure 1570. The logic may include hardware resources for interconnecting to physical interconnect hardware. The logic may include software or firmware logic for managing the interconnect. Interface controller 1534 may include a host structure interface, which may include a structure interface according to any embodiment described herein.

[0141] Node 1530 may include a memory subsystem 1540. Memory 1540 may include a memory computation resource (comp) 1542, which represents the ability of memory 1540 to perform one or more memory computations. System 1500 supports remote memory operations, such as, but not limited to, those described elsewhere herein. Therefore, node 1530 may request a remote node to perform a memory computation, wherein the data used for the computation remains local to the executing node and is not sent via structure 1570 or from memory to the structure interface. In response to the execution of the memory computation, the executing node may provide the result to the requesting node.

[0142] Processor 1532 may include one or more individual processors. Each individual processor may include a single processing unit, a multi-core processing unit, or a combination thereof. A processing unit may include a main processor, such as, but not limited to, a CPU (Central Processing Unit), a peripheral processor (such as, but not limited to, a GPU (Graphics Processing Unit)), or a combination thereof. Memory 1540 may be or include a memory device and a memory controller.

[0143] The term "memory device" can refer to different types of memory. Memory devices generally refer to volatile memory technology. Volatile memory is memory whose state (and the data stored within it) is uncertain if power is interrupted. Non-volatile memory is memory whose state is deterministic even if power is interrupted. Dynamically volatile memory can refresh the data stored in the device to maintain its state. An example of dynamically volatile memory includes DRAM (Dynamic Random Access Memory) or variations thereof, such as, but not limited to, Synchronous DRAM (SDRAM). The memory subsystem described in this article is compatible with a variety of memory technologies, such as, but not limited to, DDR3 (Double Data Rate version 3, originally released by JEDEC (Joint Electron Device Engineering Committee) on June 27, 2007, currently version 21), DDR4 (DDR version 4, initial specification released by JEDEC in September 2012), DDR4E (DDR version 4, extended version, currently under discussion by JEDEC), LPDDR3 (Low Power DDR version 3, JESD209-3B, released by JEDEC in August 2013), and LPDDR4 (Low Power Double Data Rate (LPDDR) version 4). JESD209-4 (originally released by JEDEC in August 2014), WIO2 (Wide I / O2, JESD229-2, originally released by JEDEC in August 2014), HBM (High Bandwidth DRAM, JESD235, originally released by JEDEC in October 2013), DDR5 (DDR version 5, currently under discussion by JEDEC), LPDDR5 (currently under discussion by JEDEC), HBM2 (HBM version 2, currently under discussion by JEDEC) or combinations of other memory technologies, as well as technologies derived from or extended based on such specifications.

[0144] In addition to or as an alternative to volatile memory, in one embodiment, a reference to a memory device may refer to a non-volatile memory device whose state is deterministic even when power is interrupted. In one embodiment, a non-volatile memory device is a block-addressable memory device, such as, but not limited to, NAND or NOR technology. Therefore, the memory device may also include future-generation non-volatile devices, such as, but not limited to, three-dimensional cross-point (3DXP) memory devices, other byte-addressable non-volatile memory devices, or memory devices using chalcogenide phase change materials (e.g., chalcogenide glasses). In one embodiment, the memory device may be or include multi-threshold NAND flash memory, NOR flash memory, single-level or multi-level phase change memory (PCM) or switched phase change memory (PCMS), resistive memory, nanowire memory, ferroelectric transistor random access memory (FeTRAM), magnetoresistive random access memory (MRAM) incorporating memristor technology, or spin-transfer torque (STT)-MRAM, or any combination of the foregoing, or other memories.

[0145] In at least one embodiment, system 1500 may include one or more circuits for executing an API to cause one or more storage locations of one or more scheduled instructions to be indicated to one or more users; executing an API to prevent one or more storage locations of one or more scheduled instructions from being stored; and / or executing regarding Figure 1-7 One or more of the processes illustrated or described, or otherwise performing any of the operations described above or elsewhere herein. One or more circuits may be software-configured to execute APIs to cause one or more storage locations of one or more scheduled instructions to be indicated to one or more users; to execute APIs to prevent one or more storage locations of one or more scheduled instructions from being stored; and / or to perform operations related to... Figure 1-7 One or more of the processes illustrated or described, or any of the operations performed above or elsewhere in this document.

[0146] Figure 16An accelerated processing unit 1600 according to at least one embodiment is illustrated. The accelerated processing unit 1600 may include a processor based on the CDNA architecture of AMD, Inc., Santa Clara, California, or other processors sharing at least some of the components described herein. The accelerated processing unit 1600 may include one or more accelerator complex dies (XCDs) 1604 for performing the operations described elsewhere in this document, such as, but not limited to, graphics processing and / or parallel processing and instruction-level parallel computing, including support for multiple precisions (INT8, FP8, BF16, FP16, TF32, FP32, and FP64) and sparse matrix data (i.e., sparsity). In some cases, the XCD may be referred to as a graphics computing die (GCD). The accelerated processing unit 1600 may include one or more complex computing dies (CCDs) 1606 for performing the operations described elsewhere in this document, such as, but not limited to, operations performed by a host processor. In some cases, the CCD may be referred to as a core complex or CCX, such as, but not limited to, the CCX used in AMD Ryzen processors. XCDs and CCDs can share any type of cache or memory (e.g., one or more memory cells 1602), or a cache or memory can be allocated to each XCD or CCD or group of XCDs or CCDs. For example, AMD Infinity Fabric within the package connects XCDs and CCDs to a shared AMD Infinity Cache 1608, and in some embodiments, to high-bandwidth memory (e.g., HMB3). Accelerated processing unit 1600 may include an AMD MI300a processor comprising three CPU die (or CCD) and six accelerator die (XCD) on top of four input-output dies (IODs) that may be layered on a single silicon die (e.g., via AMD Infinity Fabric) and linked together to eight high-bandwidth DRAM stacks in a ring to form a superchip. For systems using only accelerators, the AMD MI300x processor replaces the CCD with two or more XCDs.

[0147] Accelerated processing unit 1600 may include one or more input / output (I / O) interfaces. For example, XCD 1604 and CCD 1606 may coexist on one or more input-output dies (IODs) 1610, which may include one or more I / O interfaces. IOD 1610 may include any number and type of I / O interfaces (e.g., PCI, PCI expansion (“PCI-X”), PCIe, Gigabit Ethernet (“GBE”), USB, etc.). Various types of peripheral devices may be coupled to I / O interface 1670. The I / O interfaces of IOD 1610 may also be used to connect one or more accelerated processing units 1600, for example, in a server architecture.

[0148] Accelerated processing unit 1600 may include one or more memory units 1602 for storing instructions and other information for performing the operations described in other sections of this document. Memory units 1602 may include any volatile memory, such as, but not limited to, the memory types described in other sections of this document, and may include, for example, high-bandwidth memory (e.g., HMB3) or high-bandwidth DRAM. The memory associated with accelerated processing unit 1600 (e.g., memory unit 1602) may include system memory, which can be used for, for example, commands, instructions, and constants, as well as input and output. Memory unit 1602 may also include device memory, which can be used for storage and, for example, for commands, instructions, and constants, as well as input and output, as a return buffer, and for private data. Memory unit 1602 may be linked to one or more IODs 1610. In at least one embodiment, L1 cache 1620 initiates a memory hierarchy including a shared L2 cache 1628 (e.g., within an XCD). AMDInfinity Cache™ is the last-level cache (LLC) located on the active I / O die (IOD). CCD 1606 and XCD1604 can have independent or shared memory. AMD Infinity architecture and AMD Infinity Fabric™ technology enable consistent, high-throughput unification of GPU and CPU chip technologies (such as XCD, CCD, and / or CCX) with memory (such as stacked HBM3 memory) in a single device and across multiple device platforms.

[0149] like Figure 16As shown, the XCD 1604 may include a set of shared global resources 1630, which may include a hardware scheduler 1632 and an asynchronous compute engine (ACE) 1624. The ACE 1624 sends tasks (e.g., compute shader workgroups) to compute units (CUs or cores) 1634. Each of the ACEs 1624 (e.g., four) may be associated with a CU 1634 (e.g., 40 CUs), and some CUs 1634 may be disabled for yield management. CUs 1634 may have dedicated caches or shared caches (e.g., L2 caches) 1628 for consolidating all memory traffic on a single die. CU1634 may include threaded and parallel processor cores, including instruction fetching and scheduling using a scheduler (S) 1612, a matrix core unit (MCU) 1616, and a shader core (SC) 1618 (e.g., execution units for scalar, vector, and matrix data types), and a load / store pipeline with an L1 cache 1620 and a local data share (LDS) 1614. The local data share may include, for example, a sticky-note RAM with built-in arithmetic capabilities, allowing data sharing between threads in a workgroup. An instruction cache 1640 (e.g., for storing and providing instructions for performing the operations described elsewhere in this document) and a constant cache 1638 may be connected to one or more CUs and may be shared between two CUs. The matrix core 1616 can handle various data types, such as, but not limited to, INT8, FP8, FP16, BF16, and TF32 data types. Accelerated processing unit 1600 may include computation units 1634, which may be arranged in an array format, such as as a data parallel processor (DPP) array. The hyper-threaded dispatch processor 1642 can communicate with the compute unit 1634, and the command processor 1644 can read commands written by the host to memory-mapped registers in the system memory address space (not shown). When a command is completed, the command processor 1644 can send a hardware-generated interrupt to the host processor (e.g., a CCD). The memory controller 1636 can also directly access all device memories and system memory regions specified by the host. To satisfy read and write requests, the memory controller 1636 can perform the functions of a direct memory access (DMA) controller, including calculating the memory address offset based on the format of the requested data in memory. For example, one or more APIs described herein can be compiled into instructions that can be stored in the instruction cache 1640, then fetched by the instruction fetch logic in the processor 1640, decoded by the processor decoder or equivalent, scheduled (e.g., sequentially or out of order) by the scheduler or equivalent for execution, executed by the execution logic or equivalent, reordered, and then retired by the retirement logic or equivalent.The API (and / or compiled instructions including the API) can be stored in any storage device, either inside or outside the processor 1600 (e.g., in cache and / or memory). The results of the API can be stored in storage devices, either inside or outside the processor 1600, including registers, DRAM, flash memory, SRAM, cache, or other memory equivalents.

[0150] Applications may include programs running on the main processor (e.g., a CCD) and programs running on one or more XCDs (referred to as kernels). Programs can be controlled by host commands that set internal base addresses and other configuration registers, specify data fields on which the accelerator processing unit 1600 can run, invalidate and flush caches on the accelerator processing unit 1600, and cause the accelerator processing unit 1600 to begin executing a program. A kernel can be referred to as a program executed by the accelerator processing unit 1600. Kernels can execute independently on each work item or as a group of work items, referred to as a wavefront, which can execute kernels on all (e.g., 64) work items in a single pass. The computation unit 1634 may include: a scalar arithmetic logic unit (ALU) that can operate on a single value for each wavefront (shared by all work items); a vector ALU that can operate on a unique value for each work item; a local data share 1614 that allows work items within a workgroup to communicate and share data; a scalar memory (not shown) that can transfer data between the scalar general-purpose registers (SGPRs) and memory via cache; and a vector memory that can transfer data between the vector general-purpose registers (VGPRs) and memory, including sampling texture maps. Kernel control flow can be manipulated using scalar ALU instructions, which may include if / else statements, branches, and loops. Scalar ALU (SALU) and memory instructions can operate on the entire wavefront and operate on one or more SGPRs. Vector memory and ALU instructions can operate on all work items in the wavefront simultaneously.

[0151] In at least one embodiment, the acceleration processing unit 1600 may include one or more circuits for executing an API to cause one or more storage locations of one or more scheduled instructions to be indicated to one or more users; executing an API to prevent one or more storage locations of one or more scheduled instructions from being stored; and / or executing regarding Figure 1-7One or more of the processes illustrated or described, or otherwise performing any of the operations described above or elsewhere herein. One or more circuits may be software-configured to execute APIs such that one or more storage locations of one or more scheduled instructions are indicated to one or more users; execute APIs to prevent one or more storage locations of one or more scheduled instructions from being stored; and / or perform operations related to... Figure 1-7 One or more of the processes illustrated or described, or any of the operations performed above or elsewhere in this document.

[0152] Figure 17Processor 1700 is shown, including, but not limited to, a Zen architecture-based processor (e.g., Zen1, 2, 3, 4, 5, or other architectures) from AMD Inc. of Santa Clara, California, or other processors that share at least some of the components described herein. Processor 1700 includes one or more CPU dies 1702(1)-1702(N), where N is any integer greater than 1. CPU die 1702 may include any number of processor cores 1716 (e.g., for performing any operations described elsewhere herein) and any number of cache memories (e.g., for storing instructions and other information to perform any operations described elsewhere herein). For example, L2 cache unit 1718 may be coupled to processor core 1716, and processor core 1716 may share and / or be individually coupled to L2 cache unit 1718. Processor core 1716 can be coupled to L3 cache 1722 and / or a shared L3 cache, which can be the lowest level cache (LLC) 1722 used to access data and other information used by processor core 1716. One or more processor cores 1716 and one or more L2 cache units 1718 can be included in a core complex (CCX) 1720, which can include (e.g., 32MB) a shared cache (e.g., L3 cache 1722). Core complex 1720 can be manufactured onto a die (CCD or CPU die) 1702. For example, up to 12 core complexes 1720 can be configured in a processor along with 8 CPU dies 1702, providing up to 96 processor cores 1716 for processor 1700. For example, a "Zen 4c" core complex 1720 can include up to 8 cores 1716 and a shared 16MB L3 cache 1722. Two core complexes from these core complexes 1720 can be combined onto a single CPU die 1702, resulting in 16 cores per die and a total of 32MB of L3 cache per die 1722. Up to eight CPU dies 1702 can be combined with I / O units 1704 to provide up to 128 processor cores 1716 for the CPU. Up to four "Zen 4c" dies mentioned above can be combined to provide up to 64 processor cores 1716 for the CPU.

[0153] Processor 1700 may include various configurations for input / output operations, which will be further described herein. I / O unit 1704 may include one or more memory controllers 1706 capable of managing the memory usage of processor 1700 (e.g., DDR5 memory). I / O unit 1704 may include one or more SATA disk controllers for managing storage device 1712, and one or more Compute Express Link (CXL™) 1.1+ memory controllers 1714 for providing CPU-to-device and CPU-to-memory connectivity and can be flexibly assigned to specific functions during server design. I / O unit 1704 may include PCIe controller 1708 for connecting peripherals and other components connected to processor 1700. I / O unit 1704 may also include USB port 1710 for connecting to other components separate from processor 1700. CPU die 1702 may support any number of connections to I / O unit 1704, for example, one or two connections. As shown in the figure, I / O unit 1704 may include the components further described herein, and I / O unit 1704 may be an I / O die accommodating several different components. Memory controller 1706, PCIe controller 1708, USB port 1710, SATA controller 1712, and / or CXL controller 1714 may be individually integrated into any location within processor 1700, or integrated in any group or combination.

[0154] Processor 1700 may include an Infinity Fabric 1724 interconnect (which may be similar to or based on a PCIe architecture) that provides connectivity between CPUs (e.g., CPU dies 1702(1)-1702(N)), graphics processor 1726, inference engine 1732, and other components in a multi-chip architecture (e.g., security processor 1728 and I / O unit 1704). One or more AMD Infinity Fabric™ interconnects 1710 may be connected to CPU dies 1702(1)-1702(N) and used as connections between CPUs. One or more Infinity Fabric interconnects 1710 may connect each CPU die 1702 to an I / O unit 1710.

[0155] In at least one embodiment, processor 1700 may include a central processing unit (CPU) and other related hardware and software described above and further herein. Processor 1700 may also include a graphics processor 1726. Graphics processor 1726 may be used for image generation and processing, as well as other computations and operations described further herein. Graphics processor 1726 may be based on AMD's RDNA 3 or 3.5 architecture, located in Santa Clara, California. Graphics processor 1726 may include a graphics computing die (GCD) and a memory cache die (MCD). GCD may include any number of computing units (CUs) for graphics or other processing, such as operations performed by an arithmetic logic unit (ALU) described further herein. Graphics processor 1726 may include an L2 cache available for use by the computing units. MCD (not shown) may include any number of memory cells and may include a cache (e.g., an L3 cache) and a memory interface for coupling to memory (e.g., memory 1742(1)-(N), where N is an integer). Components within the graphics processor 1726 can be connected using various methods, such as using Infinity Fabric 1724 interconnects, either internally or externally to the graphics processor 1726.

[0156] Inference engine 1732 can provide neural processing capabilities to processor 1700 for computational processes used in neural networks, deep learning, and other artificial intelligence-related operations, which will be further described herein. Processor 1700 may include: a security processor 1728 for managing the security of processor 1700; a display controller 1730 for controlling the display; a system management unit 1734 for managing and operating some or all components on processor 1700; a multimedia engine 1736 for audio and video operations; a fusion controller hub 1738 for managing USB, SATA, and PCIe connections to the processor; and a sensor fusion hub 1740 for managing sensors, such as accelerometers. Processor 1700 may also include memory 1742(1)-(N), where N is any integer. Memory may include different memory types, such as LPDDR5 and / or DDR5, or other memory described elsewhere herein.

[0157] To perform the operations further described herein, processor 1700 may include an execution pipeline including a front end that may include a cache for storing instructions (e.g., an L1 cache) (not shown). A branch predictor may modify the instruction stream. Instructions may be decoded by a decoder, dispatched to a back end for execution, and renamed. For example, the instruction fetch and decode pipeline may be dispatched to integer or floating-point execution operations, which may be scheduled by a scheduler and passed to vectors and / or general-purpose registers. Floating-point multipliers and / or addition operations may be processed, and an arithmetic logic unit (ALU) may also be used to perform computations, such as arithmetic and logical operations. The output of the computation unit may be coupled to a load / store queue that may be connected to a cache, such as an L1 cache and / or an L2 cache.

[0158] With respect to processor 1700 and any components described above or elsewhere herein, one or more APIs or equivalents described herein may, for example, be compiled into instructions or equivalents (e.g., AVX-512 instructions based on a SIMD model). These instructions or equivalents may be fetched by instruction fetching logic or equivalents, decoded by processor decoder or equivalents, scheduled (e.g., sequentially or out of order) for execution by scheduler or equivalents, executed by execution logic or equivalents, reordered, and then retired by retirement logic or equivalents. APIs (and / or compiled instructions including APIs) may be stored in any storage device (e.g., cache and / or memory) internal or external to processor 1700. The results of the APIs may then be stored in storage devices internal or external to processor 1700, including registers, DRAM, flash memory, SRAM, cache, or other memory equivalents.

[0159] In at least one embodiment, processor 1700 may include one or more circuits for executing an API to cause one or more storage locations of one or more scheduled instructions to be indicated to one or more users; executing an API to prevent one or more storage locations of one or more scheduled instructions from being stored; and / or executing statements regarding... Figure 1-7 One or more of the processes illustrated or described, or otherwise performing any of the operations described above or elsewhere herein. One or more circuits may be software-configured to execute APIs to cause one or more storage locations of one or more scheduled instructions to be indicated to one or more users; to execute APIs to prevent one or more storage locations of one or more scheduled instructions from being stored; and / or to perform operations related to... Figure 1-7 One or more of the processes illustrated or described, or any of the operations performed above or elsewhere in this document.

[0160] Figure 18An example of a processing core 1800 is shown, which may implement an Arm architecture (e.g., v9.0-A) or another processor sharing at least some of the components described herein. The Neoverse™ V2 core 1800 may be implemented within a DynamIQ shared unit (DSU) cluster via a DSU-110 interconnect 1854 for connecting one or more cores, for example, for parallel processing. The Neoverse™ V2 core may be implemented as a single core within a DSU cluster configured for direct interconnection, with or without L3 cache, listener filter, or listener control unit (SCU) logic (not shown). The Neoverse™ V2 core may include a CPU bridge 1852 for connecting the core 1800 to the DSU-110 interconnect, which may also connect the core 1800 to an external memory system and the remainder of the on-chip system. The L1 instruction memory system 1802 can fetch instructions from the instruction cache 1804 and deliver the instructions (e.g., one or more APIs that can be compiled into instructions as described herein) to the instruction decoding unit 1810, for example, to perform some or all of the operations described above or elsewhere herein. The L1 instruction memory system 1802 may include the L1 instruction cache 1804 (e.g., with 64-byte cache lines), the L1 instruction translation back buffer (TLB) 1806 (e.g., natively supporting 4KB, 16KB, 64KB, and 2MB page sizes), and the macro operation cache (MOP) 1808 (e.g., a 1536-entry, 4-way skewed associated L0 MOP cache), which may include decoded and optimized instructions for higher performance. The instruction decoding unit 1810 can decode AArch64 instructions into its internal format. The register renaming unit 1812 can perform register renaming to facilitate out-of-order execution and dispatch the decoded instructions to the respective issue queues. Instruction issuing unit 1814 controls when decoded instructions are dispatched to the execution pipeline, and may include an issuing queue for storing instructions to be dispatched to the execution pipeline. Integer execution pipeline 1816 may be included in the execution pipeline and includes integer execution unit 1818 capable of performing arithmetic and logical data processing operations. Vector execution unit 1820 may be included in the execution pipeline and can execute advanced SIMD and floating-point arithmetic (FPU) 1822, execute Scalable Vector Extension (SVE) and Scalable Vector Extension 2 (SVE2) instructions 1824, and may also selectively execute cryptographic instructions 1826. Advanced SIMD may include a media and signal processing architecture that primarily adds instructions for audio, video, 3D graphics, image, and speech processing. The floating-point architecture provides support for single-precision and double-precision floating-point operations. L1 data memory system 1830 executes load and store instructions, as well as servicing memory coherence requests.The L1 data memory system 1830 may include an L1 data cache 1832 and a fully associative L1 data TLB 1834, natively supporting 4KB, 16KB, and 64KB page sizes and 2MB and 512MB block sizes. The memory management unit (MMU) 1828 provides fine-grained memory system control through a set of virtual-to-physical address mappings and memory attributes, which are stored in a translation table and saved to the TLB 1834 after address translation. The L2 memory system 1836 may include an L2 cache 1838 and can be connected to the DSU-110 1854 via an asynchronous CPU bridge 1852. The Neoverse™ V2 core 1800 supports a range of debug, test, and tracing options, including a tracing unit 1842, a tracing buffer 1840, and an embedded logic analyzer (ELA) 1848. The Neoverse™ V2 core 1800 implements the Statistical Analysis Extension (SPE) 1844, which provides a statistical view of the performance characteristics of the executed instructions. Software writers can leverage these views to optimize code for better performance. The Performance Monitoring Unit (PMU) 1846 provides a performance monitor that can be configured to collect statistics on the operation of each core and memory system. This information can be used for debugging and code analysis. The Generic Interrupt Controller (GIC) CPU interface 1850, when integrated with external allocator components, serves as a resource for supporting and managing interrupts in a clustered system. Within the cluster, there can be a CPU bridge 1852 between each Neoverse™ V2 core 1800 and the DSU-110 1854. The CPU bridge 1852 controls buffering and synchronization between the core 1800 and the DSU-110 1854. The CPU bridge 1852 can be asynchronous to allow each core 1800 to use a different frequency, power, and area implementation point. The CPU bridge 1852 can operate synchronously without affecting other interfaces, such as, but not limited to, asynchronous debug and trace interfaces.

[0161] In at least one embodiment, core 1800 may include one or more circuitry configured to execute an API to cause one or more storage locations of one or more scheduled instructions to be indicated to one or more users; execute an API to prevent one or more storage locations of one or more scheduled instructions from being stored; and / or execute statements regarding... Figure 1-7 One or more of the processes illustrated or described, or otherwise performing any of the operations described above or elsewhere herein. One or more circuits may be software-configured to execute APIs to cause one or more storage locations of one or more scheduled instructions to be indicated to one or more users; to execute APIs to prevent one or more storage locations of one or more scheduled instructions from being stored; and / or to perform operations related to... Figure 1-7 One or more of the processes illustrated or described, or any of the operations performed above or elsewhere in this document.

[0162] Figure 19 One or more chips including one or more tensor processing units (TPUs) 1900 are shown according to at least one embodiment. Figure 19 The TPU 1900 may include an application-specific integrated circuit (ASIC), for example, for performing some or all of the operations described above or elsewhere herein, such as, but not limited to, machine learning workloads that accelerate the execution of matrix operations. The TPU 1900 may be an ASIC from Alphabet Corporation, Mountain View, California. The cloud TPU includes a cloud service that enables the TPU to be used as a scalable resource for processing tasks, such as, but not limited to, machine learning workloads that can run on frameworks such as, but not limited to, TensorFlow, PyTorch, and JAX.

[0163] Chip 1900 may include any number of TPUs, which may include a tensor core 1906. Tensor core 1906 may include one or more core sequencers 1908, vector processing units (VPUs) 1910, matrix multiplication units (MXUs) 1912(A)-1914(N) (where N is any integer greater than 1), and transpose permutation units 1916. Core sequencer 1908 may fetch instructions (e.g., VLIW (Very Long Instruction Word)) from the instruction memory (Imem) of core 1906, perform scalar operations using scalar data memory (Smem) and scalar registers (Sregs) (not shown), and forward vector instructions to the vector processing units (VPUs) 1910. For example, an instruction may initiate eight operations: two scalar operations, two vector ALU operations, vector loading and storing, and queuing data into the matrix multiplication and transpose units and a pair of slots for queuing data from them. The VPU1910 can perform vector operations using a large on-chip vector memory (Vmem) and vector registers (Vregs). The VPU1910 can stream data to or from the MXU via a decoupled FIFO. The VPU1910 can collect and distribute data to the Vmem using both data-level parallelism (2D matrix and vector functional units) and instruction-level parallelism (8 operations per instruction). Large two-dimensional matrix multiplication units (MXUs) 1912(A)–1912(N) can, for example, use a systolic array to reduce area and power consumption, and use a large, software-controlled on-chip memory instead of a cache. The transpose-reduction-permute unit 1916 can perform matrix transpose, reduction, and permute operations (e.g., 128x128) on the VPU1910 channels. A high-bandwidth memory 1904 can be used for on-chip applications and can be coupled to the host queue 1902, for example, via PCIe. One or more chips 1900 can be connected together for computing. For example, one or more chips 1900 can be connected to form a torus, such as a 2D torus. Chips 1900 can also include any number (e.g., 4) of inter-core interconnect (ICI) links 1918, which can enable direct connections between chips to form a supercomputer.

[0164] For any processor in chip 1900 and any components described above or elsewhere herein, one or more APIs or equivalents described herein may, for example, be compiled into instructions or equivalents that may be fetched by instruction fetching logic or equivalents, decoded by processor decoder or equivalents, scheduled (e.g., sequentially or out of order) for execution by scheduler or equivalents, executed by execution logic or equivalents, reordered, and then retired by retirement logic or equivalents. APIs (and / or compiled instructions including APIs) may be stored in any storage device (e.g., cache and / or memory) external to or internal to any processor in chip 1900. The results of the APIs may then be stored in any storage device internal to or external to any processor in chip 1900, including registers, DRAM, flash memory, SRAM, cache, or other memory equivalents.

[0165] In at least one embodiment, chip 1900 may include one or more circuits for executing an API to cause one or more storage locations of one or more scheduled instructions to be indicated to one or more users; executing an API to prevent one or more storage locations of one or more scheduled instructions from being stored; and / or executing regarding Figure 1-7 One or more of the processes illustrated or described, or otherwise performing any of the operations described above or elsewhere herein. One or more circuits may be software-configured to execute APIs to cause one or more storage locations of one or more scheduled instructions to be indicated to one or more users; to execute APIs to prevent one or more storage locations of one or more scheduled instructions from being stored; and / or to perform operations related to... Figure 1-7 One or more of the processes illustrated or described, or any of the operations performed above or elsewhere in this document.

[0166] Figure 20A vector processor according to at least one embodiment is illustrated. The vector processor 2000 may support the RISC-V standard. The vector processor 2000 may include one or more cores 2010 (e.g., scalar units) and one or more vector processing units (VPUs) 2042 (e.g., vector units), which may, for example, perform some or all of the operations described above or elsewhere herein. The core 2010 may include an Andes Custom Extension (ACE) 2016, which can be used, for example, to deliver custom instructions to the processor 2000 via an ACP 2038. The core 2010 may include a 1-cycle multiplier and a 1-cycle instruction / data local memory (ILM / DLM) for improving parallelism by allowing simultaneous instruction fetching and data access. A memory management unit (MMU) 2024 manages system memory and cache, and provides branch execution, instruction pair issuance, L1 instruction / data cache, and local memory storage. The core 2010 may include a physical memory protection and programmable physical memory attribute unit (PMP / PPMA) 2022. Core 2010 may include a digital signal processor (DSP) 2028 and a floating-point unit (FPU) 2026, as well as a load-memory unit (LSU) 2032 for interaction with memory hierarchies (D$ 2034 and I$ 2030). Core 2010 may include a branch prediction unit 2018 and a multiplier unit 2020.

[0167] The vector processing unit (VPU) 2042 may include one or more vector function units (FU) 2046(A)-2046(N) (which may be linked together for parallel processing), a separate memory path for loading / storing RISC-V vectors (RVVs) via ACE-RVV 2048 and AndesStreaming port (ASP) 2044, and a vector load / store unit (VLSU) 2050.

[0168] The vector processor 2000 may include bus interfaces, such as, but not limited to, a cacheable L2 cache port 2056, a non-cacheable MMIO port 2054, an input-output coherence port (IOCP) 2058 for a cacheless bus master, a local memory access port for accessing the ILM / DLM 2012 (which can be coupled to SRAM 2006) and the high-bandwidth vector memory (HVM) 2036, and a shared peripheral port (SPP) 2052 for external peripherals. Other memory ports include the LM slave port AXI 2002, the HVM subordinate port AXI 2004, MEM (AXI) 2062, and AXI 2060. Trace I / F 2014 can be captured, encoded, and transmitted off-chip via Inst. Trace I / F 2008 (e.g., a record of executed processor instructions). Software tools can use Inst. Trace I / F 2008 to reconstruct the exact execution sequence of a program.

[0169] For any processor in the processor 2000 and any components described above or elsewhere herein, one or more APIs or equivalents described herein may, for example, be compiled into instructions or equivalents that may be fetched by instruction fetching logic or equivalents, decoded by processor decoder or equivalents, scheduled (e.g., sequentially or out of order) for execution by scheduler or equivalents, executed by execution logic or equivalents, reordered, and then retired by retirement logic or equivalents. The API (and / or compiled instructions including the API) may be stored in any storage device external to or internal to the processor 2000 (e.g., in cache and / or memory). The results of the API may then be stored in storage devices internal to or external to the processor 2000, including registers, DRAM, flash memory, SRAM, cache, or other memory equivalents.

[0170] In at least one embodiment, the vector processor 2000 may include one or more circuits for executing an API to cause one or more storage locations of one or more scheduled instructions to be indicated to one or more users; executing an API to prevent one or more storage locations of one or more scheduled instructions from being stored; and / or executing statements regarding... Figure 1-7 One or more of the processes illustrated or described, or otherwise performing any of the operations described above or elsewhere herein. One or more circuits may be software-configured to execute APIs to cause one or more storage locations of one or more scheduled instructions to be indicated to one or more users; to execute APIs to prevent one or more storage locations of one or more scheduled instructions from being stored; and / or to perform operations related to... Figure 1-7 One or more of the processes illustrated or described, or any of the operations described above or elsewhere in this document.

[0171] Figure 21A A schematic diagram of an example multi-core tiled processor microarchitecture is shown. Figure 21A Multi-core tiled processors in a system can include language processing processors. For example... Figure 21A As shown, each “tile” in the processor architecture is a processing element bundled together using an on-chip network (NoC) that can be used to perform some or all of the operations described above or elsewhere in this document. For example, each tile may have an instruction dispatch 2104 and integer (INT) units 2106 and floating-point (FP) units 2108, a load-memory unit (LSU) 2112 for engaging with the memory hierarchy (data cache (D$) 2110 and instruction cache (I$) 2114), and a network (NET) interface 2116 for communicating with other tiles. Some tiles in processor 2100 may include a memory controller 2102 for managing and controlling memory, as further described herein. Processor 2100 may have a functionally sliced ​​architecture. Processor 2100 may reside on an application-specific integrated circuit (ASIC). Figure 21A The layout of an ASIC can be represented. Processor 2100 may include a coprocessor designed to execute instructions for a predictive model. A predictive model refers to any model configured to make predictions based on input data. The predictive model can use a classifier for classification predictions. The predictive model can be a machine learning model, such as, but not limited to, a tensor flow model, and processor 2100 is a tensor flow processor.

[0172] The processor 2100 can employ different microarchitectures, which will Figure 21B Each tile in the process represents a separate functional unit. Conversely, the functional tiles 2124 of the processor 2100 can be aggregated into multiple functional processing units (hereinafter referred to as "slices") 2104, each corresponding to a specific functional type (e.g., FP / INT 2118, NET 2120, MEM 2122). For example, as... Figure 21B As shown, each slice may correspond to a row of functional tiles extending in a north-south direction. Furthermore, the processor 2100 may also include communication channels for carrying data between tiles in different slices, each communication channel extending horizontally in an east-west direction. Each communication channel may be connected to each slice 2104 of the processor 2100.

[0173] The slices 2104 of processor 2100 may each correspond to different functions and may include arithmetic logic slices (e.g., FP / INT 2118), channel switching slices (e.g., NET 2120), and memory slices (e.g., MEM 2122). Arithmetic logic units can perform one or more arithmetic and / or logical operations on data received via communication channels to generate output data. Examples of arithmetic logic units may be matrix multiplication units and vector multiplication units. Memory slices include memory cells that store data. Memory slices can provide data to other slices via communication channels. Memory slices can also receive data from other slices via communication channels. Channel switching slices can configurably route data from one communication channel to any other communication channel. For example, data from a first channel can be provided to a second channel via a channel switching slice. In some embodiments, a channel switching slice can be implemented as a crossbar switch. Each slice 2104 also includes its own instruction queue (not shown) for storing instructions and an instruction control unit (ICU) for controlling instruction execution. Instructions in a given instruction queue can only be executed by a tile in its associated functional slice and not by other slices of the processor.

[0174] By arranging the tiles of processor 2100 into different functional slices 2104, the on-chip instruction and control flow of processor 2100 can be separated from the data flow. For example, according to some embodiments, Figure 21B One of the arrows illustrates the instruction flow within the processor architecture. According to at least one embodiment, Figure 21B Another arrow in the diagram illustrates the data flow within the processor architecture. As shown, instruction and control flows can cross the tiles of processor 2100 in a first direction (e.g., north-south along the length of a functional slice, as indicated by the first arrow), while data flows can cross the tiles of processor 2100 in a second direction (e.g., east-west across a functional slice, as indicated by the second arrow), which is perpendicular to the first direction.

[0175] Different functional slices of processor 2100 can correspond to MEM 2122 (memory), VXM (vector execution module), MXM (matrix execution module), NIM (numerical interpretation module), and SXM (swapping and permutation module). Each slice can include N tiles, all of which can be controlled by the same instruction control unit (ICU) (not shown). Each slice can operate completely independently and can only be coordinated using barrier-like synchronization primitives or by the compiler using tractable determinism. Each tile of processor 2100 can correspond to an execution unit organized as ×M SIMD tiles. For example, each tile of the on-chip memory of processor 2100 can be organized to atomically store L-element vectors. Therefore, MEM slices with N tiles can work together to store or process large vectors (e.g., with a total of N×M elements).

[0176] Tiles within a slice can execute instructions in an "interleaved" manner, where instructions can be issued tile-by-tile within the slice over N cycle periods. Functional slices can be physically arranged on the chip to allow for efficient data flow for pipelined execution over hundreds of cycles for common patterns. The data flow can perform a single "u-turn" (direction change) corresponding to a single matrix operation before being written back to memory; in some embodiments, a particular data flow can change direction multiple times before writing the resulting data back to memory (due to multiple matrix and vector operations).

[0177] When using a processor with a function slice architecture (e.g., a TSP), the TSP compiler (not shown) generates an explicit plan of how the processor 2100 can execute programs (e.g., microprograms). The compiler can specify when each operation will be executed, which function slices will perform the work, and which STREAM registers will hold operands. The compiler can maintain a high-fidelity (cycle-accurate) model of the hardware state of the processor 2100 (e.g., the TSP) so that the microprogram can coordinate data flow.

[0178] Processor 2100 (e.g., TSP) can use a web-hosted compiler that takes a model (e.g., an ML model, such as but not limited to a TensorFlow model) as input and issues a proprietary stream of instructions for processor 2100 (e.g., TSP). The compiler is responsible for coordinating the control and data flow of the program and specifying any instruction-level parallelism by explicitly bundling instructions that can and should be executed concurrently so that they can be dispatched together. The main hardware architecture includes the architecture-visible streaming register file (STREAM), which will be described in more detail below, and acts as a conduit for operands to flow from MEM slices (e.g., SRAM) to functional slices (and vice versa).

[0179] The MEM 2122 of processor 2100 can be used as: (1) a storage for model parameters, microprograms, and data on which they operate; and (2) an on-chip network (NoC) for transferring data operands from the MEM to functional slices and returning computation results to the MEM. In some embodiments, the on-chip memory may consume approximately 75% of the chip area of ​​processor 2100. In some embodiments, the on-chip memory of the MEM tile may include SRAM instead of DRAM due to the bandwidth requirements of processor 2100. The on-chip memory capacity of processor 2100 may be determined by: (i) the number of ML models that can reside on the chip simultaneously, (ii) the size of any given model, and (iii) partitioning of large models for adaptation to a multi-chip system. In some embodiments, the MEM system of processor 2100 may provide multiple memory slices organized into two distinct hemispheres (referred to as “MEM WEST” and “MEM EAST”, respectively).

[0180] The memory slices in each hemisphere can be mirrored such that the slices are physically numbered {0,...L} in the eastern hemisphere and {L,...0} in the western hemisphere, such that memory slice 0 in each hemisphere corresponds to the slice of the VXM slice closest to the hemisphere, where each hemisphere comprises L slices. The direction of data transfer towards the chip center can be referred to as inward, while data transfer towards the outer edge of the chip (easternmost or westernmost) can be referred to as outward. Although the memory hemispheres of processor 2100 can be referred to as east and west, it is understood that other names may be used to refer to different memory hemispheres in other embodiments.

[0181] In some embodiments, streaming register files (referred to as STREAMS) transfer operands and results between the SRAM of a MEM slice of processor 2100 and the functional slice. In some embodiments, multiple MEM slices (e.g., 2 to 10 adjacent MEM slices) can be physically organized into sets. Each slice set can be located between a pair of STREAMS register files, allowing each slice to read from or write to the STREAMS registers in either direction. By placing STREAMS register files between sets of MEM slices, the number of cycles required to transfer data operands across hemispheres can be reduced (e.g., reduced by a factor corresponding to the number of slices per set). The number of slices per set can be configured based on the distance of data transfer within a single clock cycle.

[0182] for Figure 21AAny processor and any component described above or elsewhere herein, one or more APIs or equivalents described herein may, for example, be compiled into instructions or equivalents that may be fetched by instruction fetching logic or equivalents, decoded by processor decoder or equivalents, scheduled (e.g., sequentially or out of order) for execution by scheduler or equivalents, executed by execution logic or equivalents, reordered, and then retired by retirement logic or equivalents. The API (and / or compiled instructions including the API) may be stored in any storage device (e.g., in cache and / or memory) inside or outside the processor 2100. The results of the API may then be stored in storage devices inside or outside the processor 2100, including registers, DRAM, flash memory, SRAM, cache, or other memory equivalents.

[0183] In at least one embodiment, processor 2100 may include one or more circuits for executing an API to cause one or more storage locations of one or more scheduled instructions to be indicated to one or more users; executing an API to prevent one or more storage locations of one or more scheduled instructions from being stored; and / or executing statements regarding... Figure 1-7 One or more of the processes illustrated or described, or otherwise performing any of the operations described above or elsewhere herein. One or more circuits may be software-configured to execute APIs such that one or more storage locations of one or more scheduled instructions are indicated to one or more users; execute APIs to prevent one or more storage locations of one or more scheduled instructions from being stored; and / or perform operations related to... Figure 1-7 One or more of the processes illustrated or described, or any of the operations performed above or elsewhere in this document.

[0184] Software Structure The following figures illustrate, without limitation, examples of software structures for implementing at least one embodiment.

[0185] Figure 22 A software stack for a programming platform according to at least one embodiment is illustrated. The programming platform may include a platform for leveraging hardware on a computing system to accelerate computational tasks. In at least one embodiment, software developers can access the programming platform through libraries, compiler instructions, and / or extensions to programming languages. The programming platform may be CUDA, Radeon Open Computing Platform (“ROCm”), OpenCL (OpenCL™ developed by the Khronos Group), SYCL, or Intel oneAPI.

[0186] The software stack 2200 of the programming platform can provide an execution environment for the application 2201. The application 2201 may include any computer software that can be launched on the software stack 2200. The application 2201 may include artificial intelligence (“AI”) / machine learning (“ML”) applications, high-performance computing (“HPC”) applications, virtual desktop infrastructure (“VDI”), or data center workloads.

[0187] Application 2201 and software stack 2200 run on hardware 2208. Hardware 2208 may include one or more GPUs, CPUs, FPGAs, AI engines, and / or other types of computing devices supporting programming platforms. Software stack 2200 may be vendor-specific and compatible only with vendor-specific devices, such as CUDA, ROCm, OneAPI, OpenCL, or other implementations. Hardware 2208 may include a host connected to one or more devices that can be accessed via application programming interface (“API”) calls to perform computational tasks. In at least one embodiment, the devices within hardware 2208 may include GPUs, FPGAs, AI engines, or other computing devices (but may also include CPUs) and their memory, while the host within hardware 2208 may include CPUs (but may also include computing devices) and their memory. For any hardware 2208 described above or elsewhere herein, the one or more APIs described herein may, for example, be compiled into instructions that may be fetched by instruction fetching logic, decoded by a processor decoder, scheduled (e.g., sequentially or out of order) for execution by a scheduler, executed by execution logic, reordered, and then retired by retirement logic. The API (and / or compiled instructions including the API) may be stored in any storage device (e.g., cache and / or memory) internal or external to hardware 2208. The results of the API may be stored in storage devices internal or external to hardware 2208, including registers, DRAM, flash memory, SRAM, cache, or other memory. One or more APIs described herein may receive calls. One or more APIs described herein may communicate with a library or a portion of a library to perform the function described by the call. One or more APIs described herein may receive calls and communicate with a library or a portion of a library to perform the function described by the call.

[0188] The software stack 2200 of the programming platform may include multiple libraries 2203, a runtime 2205, optional drivers / interfaces 2207, and device kernel drivers 2208. Each library 2203 may include data and programming code that can be used by computer programs and utilized during software development. Library 2203 may include pre-written code and subroutines, classes, values, type specifications, configuration data, documentation, help data, and / or message templates. Library 2203 may include functionality optimized for execution on one or more types of devices. Library 2203 may include functionality for performing mathematical, deep learning, and / or other types of operations on the device. Library 2203 may be associated with corresponding APIs 2202, which may include one or more APIs for exposing the functionality implemented in library 2203. A processor (e.g., CPU, GPU) may execute, call, or otherwise use one or more APIs to determine kernel priority. For example, a first kernel (e.g., a parent kernel) may launch a second kernel (e.g., a child kernel), and the processor may use the second kernel to launch an additional kernel (e.g., a grandchild kernel) independent of the first kernel. The processor can execute APIs or call APIs from memory to support dynamic stream priorities (e.g., updating priorities when performing operations using streams). For example, when the processor executes the API, it allows the programmer to copy stream priorities from one stream to one or more other streams.

[0189] Software stack 2200 may include APIs that support dynamic stream prioritization (e.g., updating priority while performing operations on the stream), allowing programmers to set the stream's priority at any time after the stream is created. Software stack 2200 may include APIs that support dynamic stream prioritization (e.g., updating priority while performing operations on the stream), allowing programmers to obtain the stream's current priority, where the priority is one of several attributes of the stream. Software stack 2200 may include APIs that support dynamic stream prioritization (e.g., updating priority while performing operations on the stream), allowing programmers to obtain the stream's current priority as a single attribute. Software stack 2200 may include APIs that support dynamic stream prioritization (e.g., updating priority while performing operations on the stream), allowing programmers to start the kernel to perform operations on the stream at a set priority, which may be different from the stream priority. Software stack 2200 may include an API for indicating whether an object (e.g., a thread synchronization object, such as, but not limited to, a barrier) tracks whether all data movement operations of a set of threads running on the GPU have a specified state after a specified time period, wherein the specified state may be a state indicating that data has been moved and is ready for use, and is specified using expected parity values ​​as input to the API.

[0190] Software stack 2200 may include one or more APIs for updating the kernel. The processor may execute APIs or call APIs from memory to update existing APIs, thereby supporting a context-independent kernel. This allows programmers to add kernel nodes to a graph without a graph context, so that the graph context can be dynamically associated with the kernel at runtime. Software stack 2200 may include one or more APIs that allow programmers to obtain kernel identifiers and graph contexts as separate parameters from kernel nodes, thereby enabling parameter retrieval from both the kernel and the context-independent kernel. Software stack 2200 may include one or more APIs that use parallel processors (e.g., but not limited to one or more graphics processing units) to initiate task graphs (e.g., task graphs) and execute one or more task graphs (e.g., including one or more programs).

[0191] Software stack 2200 may include one or more APIs for associating one or more instructions with one or more memory sorting operations (e.g., but not limited to fence or memory barrier operations). Instructions may be associated with one or more domains, causing memory sorting operations to execute in association with one or more specific domains without interfering with instructions in other domains. APIs may indicate that a thread has reached (e.g., reached a thread synchronization barrier) or completed a certain phase of work associated with an asynchronous data movement operation on the GPU. Software stack 2200 may include one or more APIs that allow programmers to manually indicate an expected transaction count when a thread completes a certain phase of work; this transaction count can be used to update an object used to track whether all data movement operations for a set of threads have been completed.

[0192] Application 2201 can be written as source code and then compiled into executable code, as described below. Figure 23 and Figure 24 As discussed in more detail, the executable code of application 2201 can run, at least partially, within the execution environment provided by software stack 2200. During the execution of application 2201, code that needs to run on the device (rather than the host) may be encountered. In this case, runtime 2205 can be invoked to load and launch the required code on the device. Runtime 2205 can include any technically feasible runtime system capable of supporting the execution of application 2201.

[0193] Runtime 2205 can be implemented as one or more runtime libraries associated with the corresponding API (which is shown as API 2204). One or more such runtime libraries may include functions for memory management, execution control, device management, error handling, and / or synchronization, etc. Memory management functions may include functions for allocating, dealing with, and copying device memory, as well as functions for transferring data between host memory and device memory. Execution control functions may include functions for starting functions on the device (sometimes referred to as "kernels" when the function is a global function that can be called from the host) and setting attribute values ​​in buffers maintained by the runtime libraries so that a given function can be executed on the device.

[0194] Runtime libraries and corresponding APIs 2204 can be implemented in any technically feasible manner. One (or any number) of APIs can expose a set of low-level functions for fine-grained control of the device, while another (or any number) of APIs can expose a set of high-level functions for such functions. High-level runtime APIs can be built on top of low-level APIs. One or more runtime APIs can be language-specific APIs, which can be layered on top of language-independent runtime APIs.

[0195] Optional drivers or interfaces 2207 can be implemented, for example, for CUDA and ROCm implementations, which will be described further below. Optional drivers / interfaces 2207 can be associated with optional driver or interface APIs, such as, but not limited to, CUDA and / or ROCm APIs.

[0196] One or more processors disclosed in the “processing system” may execute, access, or otherwise use the software stack 2200. For example, the system-on-a-chip 900, parallel processor 1000, graphics multiprocessor 1034, processor 1100, processor 1200, accelerator 1300, neuromorphic processor 1405, supercomputer 1500, acceleration processing unit 1600, processor 1700, processor 1800, tensor processing unit 1900, processor 2000, and language processing unit 2100 may execute, use, call, or otherwise implement (e.g., by accessing memory) one or more APIs included in the software stack 2200.

[0197] Device kernel driver 2208 can be configured to facilitate communication with underlying devices. Device kernel driver 2208 can provide low-level functionality for APIs (such as, but not limited to, API 2204) and / or other software. Device kernel driver 2208 can be configured to compile intermediate representation (“IR”) code into binary code at runtime. For CUDA or other implementations (such as, but not limited to, ROCm, OneAPI, or OpenCL), device kernel driver 2208 can compile non-hardware-specific parallel thread execution (“PTX”) IR code into binary code for a specific target device at runtime (and cache the compiled binary code), which is sometimes referred to as “finalized” code. Doing so allows the finalized code to run on a target device that may not have existed when the source code was initially compiled into PTX code. Alternatively, device source code can be compiled into binary code offline without device kernel driver 2208 compiling the IR code at runtime.

[0198] Processors described elsewhere in this document (e.g., but not limited to) Figure 9-21B The processor (in the context) may include one or more circuits for executing APIs to cause one or more storage locations of one or more scheduled instructions to be indicated to one or more users; executing APIs to prevent one or more storage locations of one or more scheduled instructions from being stored; and / or executing statements regarding... Figure 1-7 One or more of the processes illustrated or described, or otherwise performing any of the operations described above or elsewhere herein. One or more circuits may be configured by software (e.g., software stack 2200) to execute APIs such that one or more storage locations of one or more scheduled instructions are indicated to one or more users; to execute APIs to prevent one or more storage locations of one or more scheduled instructions from being stored; and / or to perform operations related to... Figure 1-7 One or more of the processes illustrated or described, or any of the operations performed above or elsewhere in this document.

[0199] According to at least one embodiment, Figure 22 The software stack 2200 can execute in a CUDA implementation. The CUDA software stack 2200, on which the application 2201 can be launched, may include CUDA libraries 2203, CUDA runtime 2205, CUDA driver 2207, and device kernel driver 2208. The CUDA software stack 2200 can execute on hardware (e.g., a graphics multiprocessor 1134, which may include a CUDA-enabled GPU developed by NVIDIA Corporation, Santa Clara, California).

[0200] Application 2201, CUDA runtime 2205, and device kernel driver 2208 can perform the functions described above and elsewhere in this document. CUDA driver 2207 may include a library (libcuda.so) that implements CUDA driver API 2206. Similar to CUDA runtime API 2204 implemented by the CUDA runtime library (cudart), CUDA driver API 2206 exposes functions for memory management, execution control, device management, error handling, synchronization, and / or graphics interoperability. CUDA driver API 2206 differs from CUDA runtime API 2204 in that CUDA runtime API 2204 simplifies device code management by providing implicit initialization, context (similar to processes), and module (similar to dynamically loaded libraries) management. Compared to the high-level CUDA runtime API 2204, CUDA driver API 2206 can serve as a low-level API, providing finer-grained device control, especially in terms of context and module loading. CUDA Driver API 2206 can expose context management functions not exposed in CUDA Runtime API 2204. CUDA Driver API 2206 can also be language-agnostic, supporting technologies such as OpenCL in addition to CUDA Runtime API 2204. Furthermore, development libraries, including CUDA Runtime 2205, can be considered separate from driver components, including user-mode CUDA Driver 2207 and kernel-mode device driver 2208 (sometimes referred to as the "display" driver).

[0201] CUDA library 2203 may include mathematical libraries, deep learning libraries, parallel algorithm libraries, and / or signal / image / video processing libraries, which parallel computing applications (such as, but not limited to, application 2201) may utilize. CUDA library 2203 may include mathematical libraries, such as, but not limited to, the cuBLAS library (an implementation of the basic linear algebra subroutines (“BLAS”) for performing linear algebra operations), the cuFFT library (for computing the Fast Fourier Transform (“FFT”)), and the cuRAND library (for generating random numbers), etc. CUDA library 2203 may include deep learning libraries, such as, but not limited to, the cuDNN primitive library for deep neural networks and the TensorRT platform for high-performance deep learning inference, etc.

[0202] In at least one embodiment, the processor described elsewhere herein (e.g., but not limited to the one shown in Figure 1) Figure 9-21BThe processor (in the context) may include one or more circuits for executing APIs to cause one or more storage locations of one or more scheduled instructions to be indicated to one or more users; executing APIs to prevent one or more storage locations of one or more scheduled instructions from being stored; and / or executing statements regarding... Figure 1-7 One or more of the processes illustrated or described, or otherwise performing any of the operations described above or elsewhere herein. One or more circuits may be configured by software (e.g., software stack 2200) to execute APIs such that one or more storage locations of one or more scheduled instructions are indicated to one or more users; execute APIs to prevent one or more storage locations of one or more scheduled instructions from being stored; and / or perform operations related to... Figure 1-7 One or more of the processes illustrated or described, or any of the operations performed above or elsewhere in this document.

[0203] According to at least one embodiment, Figure 22 The software stack 2200 can execute in the ROCm implementation. An application 2201 can be launched on the ROCm software stack 2200, which includes a language runtime 2203, a system runtime 2205, a thunk 2207, and a ROCm kernel driver 2208. The ROCm software stack 2200 executes on hardware 2209, which may include a ROCm-enabled GPU developed by AMD Inc. of Santa Clara, California.

[0204] Application 2201 is executable and can be combined with the above. Figure 22 Similar functionality to that discussed. Furthermore, the language runtime 2203 and system runtime 2205 can execute functions related to the above. Figure 22The runtime discussed is similar in functionality to 2205. The difference between the language runtime 2203 and the system runtime 2205 is that the system runtime 2205 is a language-independent runtime that implements the ROCr System Runtime API 2204 and uses the Heterogeneous System Architecture (“HSA”) runtime API. The HSA runtime API may include a streamlined user-mode API that exposes interfaces for accessing and interacting with the AMD GPU, including functions for memory management, execution control via kernel architecture dispatch, error handling, system and agent information, and runtime initialization and shutdown. Unlike the system runtime 2205, the language runtime 2203 may be an implementation of the language-specific runtime API 2202, which sits in a layer above the ROCr System Runtime API 2204. The language runtime API may include the Heterogeneous Computing Portable Interface (“HIP”) language runtime API, the Heterogeneous Computing Compiler (“HCC”) language runtime API, or the OpenCL API, etc. HIP is an extension of the C++ programming language, offering a functionally similar version to the CUDA mechanism. Furthermore, the HIP runtime API can include features related to the above. Figure 22 The discussion covers functions similar to the CUDA runtime API, such as, but not limited to, memory management, execution control, device management, error handling, and synchronization.

[0205] The Thunk (ROCt) 2207 can be interface 2206, which is used to interact with the underlying ROCm driver 2208. The ROCm driver 2208 can be the ROCk driver, a combination of the AMD GPU driver and the HSA core driver (amdkfd). The AMD GPU driver can be a device core driver developed by AMD for GPUs, performing functions similar to those described above. Figure 22 The device kernel driver discussed is similar to the 2209. An HSA kernel driver can be a driver that allows different types of processors to share system resources more efficiently through hardware features.

[0206] Various libraries (not shown) may be included in the ROCm software stack 2200 on top of the language runtime 2203, and provide integration with the above. Figure 22 The discussion focuses on CUDA library 2203 and similar functionality. Various libraries can include mathematical libraries, deep learning libraries, and / or other libraries, such as, but not limited to, the hipBLAS library which implements functionality similar to CUDA cuBLAS, the rocFFT library for computing FFTs similar to CUDA cuFFT, etc.

[0207] Processors described elsewhere in this document (e.g., but not limited to) Figure 9-21BThe processor (in the context) may include one or more circuits for executing APIs to cause one or more storage locations of one or more scheduled instructions to be indicated to one or more users; executing APIs to prevent one or more storage locations of one or more scheduled instructions from being stored; and / or executing statements regarding... Figure 1-7 One or more of the processes illustrated or described, or otherwise performing any of the operations described above or elsewhere herein. One or more circuits may be configured by software (e.g., software stack 2200) to execute APIs such that one or more storage locations of one or more scheduled instructions are indicated to one or more users; execute APIs to prevent one or more storage locations of one or more scheduled instructions from being stored; and / or perform operations related to... Figure 1-7 One or more of the processes illustrated or described, or any of the operations performed above or elsewhere in this document.

[0208] According to at least one embodiment, Figure 22 The software stack 2200 can execute in an OpenCL implementation. The OpenCL software stack 2200, on which the application 2201 can be launched, may include the OpenCL framework 2203, the OpenCL runtime 2205, and a driver 2208. The OpenCL software stack 2200 can execute on non-vendor-specific hardware 2209. Because devices developed by different vendors support OpenCL, specific OpenCL drivers may be required for interoperability with the hardware of those vendors.

[0209] Application 2201, OpenCL runtime 2205, device kernel driver 2208, and hardware 2209 can execute in combination with the above. Figure 22 Other implementations of the discussed application 2201, runtime 2205, device kernel driver 2208, and hardware 2209 offer similar functionality. Application 2201 may also include an OpenCL kernel (not shown), whose code will execute on the device.

[0210] OpenCL can define a "platform" that allows a host to control devices connected to it. The OpenCL framework provides platform-level APIs and runtime APIs, shown as Platform API 2202 and Runtime API 2204, respectively. Runtime API 2204 uses contexts to manage kernel execution on devices. Each identified device can be associated with a corresponding context, which Runtime API 2204 uses to manage the device's command queue, program objects, kernel objects, shared memory objects, and so on. Platform API 2202 exposes functions that allow the use of device contexts to select and initialize devices, submit work to devices via command queues, and enable data transfer with devices. In addition, the OpenCL framework provides various built-in functions (not shown), including mathematical functions, relational functions, and image processing functions.

[0211] The OpenCL framework 2203 may also include a compiler (not shown). Source code can be compiled offline before application execution or online during application execution. Unlike CUDA and ROCm, OpenCL applications can be compiled online by a compiler, which represents any number of compilers that can be used to compile source code and / or IR code (e.g., but not limited to Standard Portable Intermediate Representation (“SPIR-V”) code) into binary code. Alternatively, OpenCL applications can be compiled offline before execution.

[0212] In at least one embodiment, the processor described elsewhere herein (e.g., but not limited to) Figure 9-21B The processor (in the context) may include one or more circuits for executing APIs to cause one or more storage locations of one or more scheduled instructions to be indicated to one or more users; executing APIs to prevent one or more storage locations of one or more scheduled instructions from being stored; and / or executing statements regarding... Figure 1-7 One or more of the processes illustrated or described, or otherwise performing any of the operations described above or elsewhere herein. One or more circuits may be configured by software (e.g., software stack 2200) to execute APIs such that one or more storage locations of one or more scheduled instructions are indicated to one or more users; execute APIs to prevent one or more storage locations of one or more scheduled instructions from being stored; and / or perform operations related to... Figure 1-7 One or more of the processes illustrated or described, or any of the operations performed above or elsewhere in this document.

[0213] According to at least one embodiment, the software may be supported by a programming platform configured to support various programming models, middleware and / or libraries, and frameworks that the application may rely on. The application may be an AI / ML application implemented using, for example, deep learning frameworks (such as, but not limited to, MXNet, PyTorch, or TensorFlow), which may rely on libraries such as, but not limited to, cuDNN, the NVIDIA Collective Communication Library (“NCCL”), and / or the NVIDIA Developer Data Loading Library (“DALI”) CUDA library to provide accelerated computation on the underlying hardware.

[0214] The programming platform can be a combination of the above. Figure 22 The platform described is one of CUDA, ROCm, or OpenCL. The programming platform can support various programming models, which can be abstractions of the underlying computing system that allow the expression of algorithms and data structures. Programming models can expose features of the underlying hardware to improve performance. Programming models may include CUDA, HIP, OpenCL, C++ Accelerated Massive Parallelism (“C++AMP”), Open Multiprocessing (“OpenMP”), Open Accelerators (“OpenACC”), and / or Vulkan Compute.

[0215] Libraries and / or middleware can provide abstract implementations of programming models. Such libraries may include data and programming code that computer programs can use and leverage during software development. Such middleware may include software that provides services to applications beyond those offered by the programming platform. Libraries and / or middleware may include cuBLAS, cuFFT, cuRAND, and other CUDA libraries, or rocBLAS, rocFFT, rocRAND, and other ROCm libraries. Furthermore, libraries and / or middleware may include the NCCL and ROCm communication collection library (“RCCL”) libraries that provide communication routines for GPUs, the MIOpen library for accelerating deep learning, and / or the Eigen library for linear algebra, matrix and vector operations, geometric transformations, numerical solvers, and related algorithms.

[0216] Application frameworks may depend on libraries and / or middleware. Each application framework can be a software framework that provides a standard structure for implementing application software. Returning to the AI / ML example discussed above, AI / ML applications can be implemented using frameworks such as, but not limited to, deep learning frameworks like Caffe, Caffe2, TensorFlow, Keras, PyTorch, or MxNet.

[0217] In at least one embodiment, the processor described elsewhere herein (e.g., but not limited to) Figure 9-21BThe processor (in the context) may include one or more circuits for executing APIs to cause one or more storage locations of one or more scheduled instructions to be indicated to one or more users; executing APIs to prevent one or more storage locations of one or more scheduled instructions from being stored; and / or executing statements regarding... Figure 1-7 One or more of the processes illustrated or described herein, or otherwise performing any of the operations described above or elsewhere herein. One or more circuits may be configured by software (such as the programming platform described herein) to execute APIs such that one or more storage locations of one or more scheduled instructions are indicated to one or more users; execute APIs to prevent one or more storage locations of one or more scheduled instructions from being stored; and / or perform operations related to... Figure 1-7 One or more of the processes illustrated or described, or any of the operations performed above or elsewhere in this document.

[0218] Figure 23 The method for using at least one embodiment in the above is shown. Figure 22 Compiled code executed on one of the programming platforms shown. Compiler 2301 is configured to receive source code 2300, compile source code 2300, and output executable file 2310. Compiler 2301 can be configured to convert source code 2300 into host executable code 2307 for execution on a host and device executable code 2308 for execution on a device. Source code 2300 can be compiled offline before executing the application or can be compiled online during application execution. Source code 2300 can include code in any programming language supported by compiler 2301, such as, but not limited to, C++, C, Fortran, etc. Source code 2300 can be included in a single source file containing both host code and device code, indicating the location of the device code. The single source file can be a .cu file including CUDA code, or a .hip.cpp file including HIP code, or a file in other formats including both host code and device code. Alternatively, source code 2300 can include multiple source code files instead of a single source file, with host code and device code separated into these files. Compiler 2301 includes or can access one or more libraries to identify API call sequences to execute a single fused API, wherein the single fused API is a combination of two or more APIs. In at least one embodiment, compiler 2301 may be an NVIDIA CUDA compiler (“NVCC”) for compiling CUDA code in a .cu file, or an HCC compiler for compiling HIP code in a .hip.cpp file, or other compilers.

[0219] Compiler 2301 can be configured to compile source code 2300 into host executable code 2307 for execution on a host and device executable code 2308 for execution on a device. The operations performed by compiler 2301 include resolving source code 2300 into an abstract system tree (AST), performing optimizations, and generating executable code. When source code 2300 comprises a single source file, compiler 2301 can separate the device code from the host code in that single source file, compile the device code and host code into device executable code 2308 and host executable code 2307 respectively, and link the device executable code 2308 and host executable code 2307 together to form a single file.

[0220] Compiler 2301 may include compiler front-end 2302, host compiler 2305, device compiler 2306, and linker 2309. Compiler front-end 2302 may be configured to separate device code 2304 from host code 2303 in source code 2300. In at least one embodiment, device code 2304 may be compiled by device compiler 2306 into device executable code 2308, which may include binary code or IR code as described above. Separately, host code 2303 may be compiled by host compiler 2305 into host executable code 2307. For other compilers such as NVCC (e.g., but not limited to oneAPI, ROCm, and OpenCL compilers), host compiler 2305 may be a general-purpose C / C++ compiler that outputs native object code, while device compiler 2306 may be a low-level virtual machine (“LLVM”) based compiler that forks the LLVM compiler infrastructure and outputs PTX code or binary code. For HCC, both host compiler 2305 and device compiler 2306 may be LLVM based compilers that output object binary code.

[0221] After compiling source code 2300 into host executable code 2307 and device executable code 2308, linker 2309 can link the host executable code 2307 and device executable code 2308 together to form executable file 2310. The host's native object code and the device's PTX or binary code can be linked together in an executable and linkable format (“ELF”) file, a container format for storing object code. Host executable code 2307 and device executable code 2308 can take any suitable format, such as, but not limited to, binary code and / or IR code. In at least one embodiment, for CUDA, host executable code 2307 may include native object code, while device executable code 2308 may include code in a PTX intermediate representation. In at least one embodiment, for ROCm, both host executable code 2307 and device executable code 2308 can include object binary code. Other implementations (e.g., but not limited to oneAPI, OpenCL) are considered and can be performed similarly to the CUDA and ROCm implementations described above.

[0222] Source code 2300 can be transformed before compilation. The source code is passed through a transformation tool (not shown), which transforms source code 2300 into transformed source code. Compiler 2301 can be used to compile the transformed source code into host executable code 2307 and device executable code 2308, a process similar to compiler 2301 compiling source code 2300 into host executable code 2307 and device executable code 2308, as described above. Figure 23 The discussion.

[0223] The transformations performed by the transformation tools can be used to port Source Code 2300 to environments different from the environment it was originally intended to run in. The transformation tools may include a HIP converter, which is used to "hipify" CUDA code intended for the CUDA platform into HIP code that can be compiled and executed on the ROCm platform. The transformation of Source Code 2300 may include parsing Source Code 2300 and translating calls to APIs provided by one programming model (e.g., CUDA) into corresponding calls to APIs provided by another programming model (e.g., HIP), as described below. Figure 24 Let's discuss this in more detail. Returning to the example of HIP-enhanced CUDA code, calls to the CUDA runtime API, CUDA driver API, and / or CUDA libraries can be translated into corresponding HIP API calls. The automatic conversion performed by conversion tool 2301 may sometimes be incomplete, requiring additional manual intervention to fully port the source code 2300.

[0224] One or more techniques described herein can utilize various methods for converting one type of code into another. For example, compiler 2301 or other compilers described herein can convert a high-level language (e.g., source code abstracting hardware) into a low-level language (e.g., machine code or intermediate representation). Source code can be scanned, parsed, converted into an abstract syntax tree (API), which is then semantically analyzed, converted into intermediate code, and then into machine code or assembly language. Compiler 2301 or other compilers described herein may include a transpiler that can, for example, convert one type of source code into another type of source code, or convert one type of machine code into another type of machine code. Source code can be parsed and converted into an API, which can then be converted into an intermediate model, which can be converted into an API of the target language, and code can be generated. Compiler 2301 or other compilers described herein can be used to achieve interoperability between different device architectures. For example, an application for one platform (e.g., a CUDA application) can be compiled into code for implementation on another platform (e.g., an AMD processor, an Intel processor, or another processor). Source code 2300 may include source code for one platform (e.g., CUDA). Compiler 2301 can compile source code 2300 into an executable file 2310 that can be used by another platform (e.g., AMD or Intel). Programming toolkits can allow applications intended for one platform (e.g., CUDA) to be compiled (e.g., natively compiled) for another platform (e.g., AMD or Intel). For example, the GPGPU programming toolkit can allow CUDA applications to be compiled natively for AMD GPUs. Programs (e.g., CUDA programs) or their build systems do not require modification or conversion to another language before being compiled into code for another platform. The compiler can accept the same command-line options and programming cognates (e.g., CUDA cognates) as another compiler (e.g., nvcc for CUDA), serving as a direct replacement for emulation toolkit (e.g., NVIDIA CUDA Toolkit) installations, so existing build tools and scripts (e.g., cmake) work without further modification. In at least one embodiment, an nvcc-cognate CUDA can be compiled for AMD GPUs (including PTX asm) using an nvcc-compatible compiler. Implementations of the CUDA runtime and driver APIs for AMD GPUs can be used. Libraries (e.g., open-source wrapper libraries) can provide APIs by delegating to the corresponding ROCm libraries, such as the "CUDA-X" API. Example implementations include SCALE from Spectral Compute in London, UK.SCALE allows programs written in CUDA languages ​​to be directly compiled into low-level languages ​​(e.g., machine code) for AMD GPUs. SCALE can create one or more directories that can be used to emulate the NVIDIA CUDA Toolkit (from the build system's perspective) by indicating to the build system that the CUDA installation path is provided by SCALE, not by NVIDIA. Additional implementations may include a Clang compiler that provides language front-ends and tooling infrastructure for languages ​​in the C language family (C, C++, Objective C / C++, OpenCL, CUDA, and RenderScript). In at least one embodiment, the compilers and / or transpilers described herein (e.g., but not limited to compiler 2301, compiler 2305, and / or compiler 2306) may include one or more circuits for compiling code (e.g., CUDA, HIP, OpenCL, OneAPI, or others) to execute APIs such that one or more storage locations of one or more scheduled instructions are indicated to one or more users; to execute APIs to prevent one or more storage locations of one or more scheduled instructions from being stored; and / or to perform actions related to... Figure 1-7 One or more of the processes illustrated or described herein, or any of the operations described above or elsewhere herein, may be performed. In at least one embodiment, the compiler and / or translator described herein (e.g., but not limited to compiler 2301, compiler 2305, and / or compiler 2306) may include one or more circuitry for translating code (e.g., CUDA source code) into one or more other types of code (e.g., machine code for CUDA and / or other platforms such as AMD or Intel processors), for executing APIs to cause one or more storage locations of one or more scheduled instructions to be indicated to one or more users; for executing APIs to prevent one or more storage locations of one or more scheduled instructions from being stored; and / or for executing... Figure 1-7 One or more of the processes illustrated or described, or any operation performed in any way as described above or elsewhere in this document.

[0225] Figure 24A system 2400 is shown, configured to compile and execute CUDA source code 2410 using different types of processing units according to at least one embodiment. The system 2400 includes CUDA source code 2410, CUDA compiler 2450, host executable code 2470 (1), host executable code 2470 (2), CUDA device executable code 2484, CPU 2490, CUDA-enabled GPU 2494, GPU 2492, CUDA to HIP conversion tool 2420, HIP source code 2430, HIP compiler driver 2440, HCC 2460, and HCC device executable code 2482.

[0226] CUDA source code 2410 can be a collection of human-readable code in the CUDA programming language. The CUDA programming language can be an extension of the C++ programming language, including mechanisms for defining device code and distinguishing between device code and host code. Device code can include source code that, after compilation, can be executed in parallel on a device. A device can be a processor optimized for parallel instruction processing, such as, but not limited to, a CUDA-enabled GPU 2490, GPU 2492, or other GPGPUs. Host code is source code that, after compilation, can be executed on a host machine. A host machine is a processor optimized for sequential instruction processing, such as, but not limited to, a CPU 2490.

[0227] CUDA source code 2410 may include any number (including zero) of global functions 2412, any number (including zero) of device functions 2414, any number (including zero) of host functions 2416, and any number (including zero) of host / device functions 2418. Global functions 2412, device functions 2414, host functions 2416, and host / device functions 2418 can be mixed within CUDA source code 2410. Each global function 2412 can be executed on the device and called from the host. Therefore, one or more global functions 2412 can act as entry points for the device. Each global function 2412 can be a kernel. In a technique called dynamic parallelism, one or more global functions 2412 can define a kernel that can be executed on the device and called from the device. During execution, the kernel can be executed in parallel N times by N different threads on the device (where N is any positive integer).

[0228] Each device function 2414 can be executed on a device and can only be called from that device. Each host function 2416 can be executed on a host and can only be called from that host. Each host / device function 2416 can define a host version of a function that can be executed on a host and called only from that host, and a device version of a function that can be executed on a device and called only from that device.

[0229] CUDA source code 2410 can also include any number of calls to any number of functions, which can be defined through CUDA runtime API 2402. CUDA runtime API 2402 can include any number of functions that execute on the host to allocate and deallocate device memory, transfer data between host memory and device memory, manage systems with multiple devices, etc. CUDA source code 2410 can also include any number of calls to any number of functions, which can be specified in any number of other CUDA APIs. CUDA APIs can be any API designed for use by CUDA code. CUDA APIs can include CUDA runtime API 2402, CUDA driver APIs, APIs for any number of CUDA libraries, etc., including any APIs described elsewhere in this document. Compared to CUDA runtime API 2402, CUDA driver APIs can be lower-level APIs but can provide more fine-grained control over devices. Examples of CUDA libraries include cuBLAS, cuFFT, cURAND, cuDNN, etc.

[0230] CUDA compiler 2450 can compile input CUDA code (e.g., CUDA source code 2410) to generate host executable code 2470(1) and CUDA device executable code 2484. CUDA compiler 2450 can be, but is not limited to, NVCC. Host executable code 2470(1) can be a compiled version of the host code included in the input source code, which can be executed on CPU 2490. CPU 2490 can be any processor optimized for sequential instruction processing.

[0231] CUDA device executable code 2484 may be a compiled version of the device code included in the input source code, which can be executed on a CUDA-enabled GPU 2494. CUDA device executable code 2484 may include binary code. CUDA device executable code 2484 may include IR code (such as, but not limited to, PTX code), which is further compiled at runtime by the device driver into binary code for a specific target device (e.g., a CUDA-enabled GPU 2494). CUDA-enabled GPU 2494 may include any processor optimized for parallel instruction processing and supporting CUDA. CUDA-enabled GPU 2494 may be developed by NVIDIA Corporation, located in Santa Clara, California.

[0232] The CUDA to HIP conversion tool 2420 can be configured to convert CUDA source code 2410 into functionally similar HIP source code 2430. The HIP source code 2430 may include a collection of human-readable code written in the HIP programming language. The HIP code may include human-readable code written in the HIP programming language. The HIP programming language may include extensions to the C++ programming language that include functionally similar versions of the CUDA mechanisms for defining device code and distinguishing it from host code. The HIP programming language may include a subset of the functionality of the CUDA programming language. For example, the HIP programming language may include mechanisms for defining global functions 2412, but such HIP programming languages ​​may lack support for dynamic parallelism, so the global function 2412 defined in the HIP code may only be called from the host.

[0233] HIP source code 2430 may include any number (including zero) of global functions 2412, any number (including zero) of device functions 2414, any number (including zero) of host functions 2416, and any number (including zero) of host / device functions 2418. HIP source code 2430 may also include any number of calls to any number of functions specified in HIP runtime API 2432. HIP runtime API 2432 may include functionally similar versions of a subset of functions contained in CUDA runtime API 2402. HIP source code 2430 may also include any number of calls to any number of functions specified in any number of other HIP APIs. HIP APIs may be any API designed for use by HIP code and / or ROCm. HIP APIs may include HIP runtime API 2432, HIP driver APIs, APIs for any number of HIP libraries, APIs for any number of ROCm libraries, etc.

[0234] The CUDA to HIP conversion tool 2420 can convert each kernel call in CUDA code from CUDA syntax to HIP syntax, and can convert any number of other CUDA calls in CUDA code into any number of other functionally similar HIP calls. CUDA calls can include calls to functions specified in the CUDA API, and HIP calls can include calls to functions specified in the HIP API. The CUDA to HIP conversion tool 2420 can convert any number of calls to functions specified in the CUDA runtime API 2402 into any number of calls to functions specified in the HIP runtime API 2432.

[0235] The CUDA to HIP conversion tool 2420 may include a tool called hipify-perl, which performs a text-based conversion process. The CUDA to HIP conversion tool 2420 may also include a tool called hipify-clang, which, compared to hipify-perl, performs a more complex and robust conversion process, including parsing the CUDA code using clang (a compiler front-end) and then converting the resulting symbols. Converting CUDA code to HIP code may include modifications beyond those performed by the CUDA to HIP conversion tool 2420 (e.g., manual editing).

[0236] HIP compiler driver 2440 may include a front-end that identifies target device 2446 and then configures a compiler compatible with target device 2446 to compile HIP source code 2430. Target device 2446 may include a processor optimized for parallel instruction processing. HIP compiler driver 2440 may identify target device 2446 in any technically feasible manner.

[0237] If the target device 2446 is CUDA compatible (e.g., a CUDA-enabled GPU 2494), the HIP compiler driver 2440 can generate HIP / NVCC compilation command 2442. The HIP / NVCC compilation command 2442 can configure the CUDA compiler 2450 to compile HIP source code 2430 using a HIP-to-CUDA translation header and the CUDA runtime library. In response to the HIP / NVCC compilation command 2442, the CUDA compiler 2450 can generate host executable code 2470(1) and CUDA device executable code 2484.

[0238] If the target device 2446 is incompatible with CUDA, the HIP compiler driver 2440 can generate HIP / HCC compilation command 2444. HIP / HCC compilation command 2444 can configure HCC 2460 to compile HIP source code 2430 using the HCC header and HIP / HCC runtime library. In response to HIP / HCC compilation command 2444, HCC 2460 can generate host executable code 2470(2) and HCC device executable code 2482. HCC device executable code 2482 can be a compiled version of the device code included in HIP source code 2430, which can be executed on GPU 2492. GPU 2492 can be any processor optimized for parallel instruction processing, incompatible with CUDA, and compatible with HCC. GPU 2492 can be developed by AMD Inc., located in Santa Clara, California. GPU 2492 can include GPU 2492 that does not support CUDA.

[0239] For illustrative purposes only, Figure 24 Three different flows, which can be implemented in at least one embodiment, are described for compiling CUDA source code 2410 for execution on CPU 2490 and various devices. A direct CUDA stream can compile CUDA source code 2410 for execution on CPU 2490 and CUDA-enabled GPU 2494 without converting CUDA source code 2410 to HIP source code 2430. An indirect CUDA stream can convert CUDA source code 2410 to HIP source code 2430 and then compile HIP source code 2430 for execution on CPU 2490 and CUDA-enabled GPU 2494. A CUDA / HCC stream can convert CUDA source code 2410 to HIP source code 2430 and then compile HIP source code 2430 for execution on CPU 2490 and GPU 2492.

[0240] The achievable direct CUDA stream is represented by dashed lines and a series of bubbles labeled A1-A3. As shown in bubble A1, CUDA compiler 2450 can receive CUDA source code 2410 and CUDA compilation command 2448, which configures CUDA compiler 2450 to compile CUDA source code 2410. The CUDA source code 2410 available for direct CUDA stream can be written in a CUDA programming language based on a programming language other than C++ (e.g., C, Fortran, Python, Java, etc.). In response to CUDA compilation command 2448, CUDA compiler 2450 can generate host executable code 2470(1) and CUDA device executable code 2484 (shown in bubble A2). As shown in bubble A3, host executable code 2470(1) and CUDA device executable code 2484 can be executed on CPU 2490 and CUDA-enabled GPU 2494, respectively. CUDA device executable code 2484 may include binary code. CUDA device executable code 2484 may include PTX code and may be further compiled at runtime into binary code for a specific target device.

[0241] The achievable indirect CUDA streams are represented by dashed lines and a series of bubbles labeled B1-B6. As shown in bubble B1, the CUDA-to-HIP conversion tool 2420 can receive CUDA source code 2410. As shown in bubble B2, the CUDA-to-HIP conversion tool 2420 can convert CUDA source code 2410 into HIP source code 2430. As shown in bubble B3, the HIP compiler driver 2440 can receive HIP source code 2430 and determine that the target device 2446 supports CUDA.

[0242] As indicated by bubble B4, the HIP compiler driver 2440 can generate HIP / NVCC compilation command 2442 and transfer HIP / NVCC compilation command 2442 and HIP source code 2430 to CUDA compiler 2450. HIP / NVCC compilation command 2442 can configure CUDA compiler 2450 to compile HIP source code 2430 using HIP to CUDA translation header and CUDA runtime library. HIP to CUDA translation header can convert any number of mechanisms (e.g., functions) specified in any number of HIP APIs to any number of mechanisms specified in any number of CUDA APIs. CUDA compiler 2450 can combine HIP to CUDA translation header and CUDA runtime library corresponding to CUDA runtime API 2402 to generate host executable code 2470(1) and CUDA device executable code 2484. In response to HIP / NVCC compilation command 2442, CUDA compiler 2450 can generate host executable code 2470(1) and CUDA device executable code 2484 (represented by bubble B5). As shown in bubble B6, host executable code 2470(1) and CUDA device executable code 2484 can be executed on CPU 2490 and CUDA-enabled GPU 2494, respectively. CUDA device executable code 2484 may include binary code. CUDA device executable code 2484 may include PTX code and can be further compiled at runtime into binary code for a specific target device.

[0243] The implementable CUDA / HCC streams are represented by solid lines and a series of bubbles labeled C1-C6. As shown in bubble C1, the CUDA-to-HIP conversion tool 2420 can receive CUDA source code 2410. As shown in bubble C2, the CUDA-to-HIP conversion tool 2420 can convert CUDA source code 2410 into HIP source code 2430. As shown in bubble C3, the HIP compiler driver 2440 can receive HIP source code 2430 and can determine that the target device 2446 does not support CUDA.

[0244] HIP compiler driver 2440 can generate HIP / HCC compilation command 2444 and transfer HIP / HCC compilation command 2444 and HIP source code 2430 to HCC 2460 (shown as bubble C4). HIP / HCC compilation command 2444 can configure HCC 2460 to compile HIP source code 2430 using HCC header files and HIP / HCC runtime library. HIP / HCC runtime library can correspond to HIP runtime API 2432. HCC header can include any number and type of HIP and HCC interoperability mechanisms. In response to HIP / HCC compilation command 2444, HCC 2460 can generate host executable code 2470 (2) and HCC device executable code 2482 (shown as bubble C5). As shown in the bubble labeled C6, host executable code 2470(2) and HCC device executable code 2482 can be executed on CPU 2490 and GPU 2492, respectively.

[0245] After converting CUDA source code 2410 to HIP source code 2430, executable code for a CUDA-enabled GPU 2494 or GPU 2492 can be generated using HIP compiler driver 2440 without re-executing the CUDA to HIP conversion tool 2420. CUDA to HIP conversion tool 2420 can convert CUDA source code 2410 to HIP source code 2430 and then store the HIP source code 2430 in memory. HIP compiler driver 2440 can then configure HCC 2460 to generate host executable code 2470 (2) and HCC device executable code 2482 based on the HIP source code 2430. In at least one embodiment, HIP compiler driver 2440 subsequently configures CUDA compiler 2450 to generate host executable code 2470 (1) and CUDA device executable code 2484 based on the stored HIP source code 2430.

[0246] According to at least one embodiment, the example kernel may be derived from... Figure 24 The CUDA to HIP conversion tool 2420 performs the conversion. The CUDA source code 2410 divides the overall problem that a given kernel is designed to solve into relatively coarse subproblems, which can be solved independently using thread blocks. Each thread block contains any number of threads. Each subproblem can be divided into relatively fine pieces, which can be solved collaboratively in parallel by threads within the thread block. Threads within a thread block can cooperate by sharing data via shared memory and synchronizing execution to coordinate memory access.

[0247] CUDA source code 2410 can organize thread blocks associated with a given kernel into a one-dimensional, two-dimensional, or three-dimensional thread block grid. Each thread block contains any number of threads, and the grid contains any number of thread blocks.

[0248] A kernel can be a function defined in device code using the "__global__" declaration specifier. The dimensions of the raster of kernels and their associated streams that execute a given kernel call can be specified using the CUDA kernel startup syntax. The CUDA kernel startup syntax is specified as "KernelName <<<GridSize,BlockSize,SharedMemorySize,Stream> >> (KernelArguments);. The execution configuration syntax can include a "<<<...>>>" structure inserted between the kernel name ("KernelName") and the parenthesized list of kernel arguments ("KernelArguments"). The CUDA kernel boot syntax can include the CUDA boot function syntax, instead of the execution configuration syntax.

[0249] "GridSize" can be of type dim3 and specifies the dimensions and size of the grid. The dim3 type can be a CUDA-defined structure containing unsigned integers x, y, and z. If z is not specified, it defaults to 1. If y is not specified, it defaults to 1. The number of thread blocks in the grid can be equal to the product of GridSize.x, GridSize.y, and GridSize.z. "BlockSize" can be of type dim3 and specifies the dimensions and size of each thread block. The number of threads per thread block can be equal to the product of BlockSize.x, BlockSize.y, and BlockSize.z. Each thread executing the kernel can be assigned a unique thread ID, which can be accessed in the kernel via built-in variables such as "threadIdx".

[0250] Regarding the CUDA kernel startup syntax, "SharedMemorySize" is an optional argument that, in addition to statically allocated memory, can specify the number of bytes in shared memory dynamically allocated per thread block for a given kernel call. The default value for SharedMemorySize is zero. Regarding the CUDA kernel startup syntax, "Stream" is also an optional argument that specifies the associated stream; the default value is zero to specify the default stream. A stream can be a sequence of commands executed sequentially (which may be emitted by different host threads). Different streams can execute commands out of order or concurrently.

[0251] CUDA source code 2410 may include the kernel definition and main function of the example kernel "MatAdd". The main function may be host code that executes on the host machine and includes a kernel call that causes the MatAdd kernel to execute on the device. The MatAdd kernel can add two NxN matrices A and B, where N is a positive integer, and store the result in matrix C. The main function can define the threadsPerBlock variable as 16x16 and the numBlocks variable as N / 16 x N / 16. Then, the main function can specify the kernel call "MatAdd<<<numBlocks,threadsPerBlock> >> (A, B, C);”. According to the CUDA kernel startup syntax, the kernel MatAdd can be executed using a thread block grid of dimensions N / 16 x N / 16, where each thread block is 16x16. Each thread block can contain 256 threads, and a grid with enough thread blocks can be created so that there is one thread per matrix element, and each thread in such a grid can execute the kernel MatAdd to perform a pairwise addition.

[0252] When converting CUDA source code 2410 to HIP source code 2430, the CUDA-to-HIP conversion tool 2420 can convert each kernel call in the CUDA source code 2410 from the CUDA kernel startup syntax to the HIP kernel startup syntax, and can convert any number of other CUDA calls in the source code 2410 into any number of other functionally similar HIP calls. The HIP kernel startup syntax can be specified as "hipLaunchKernelGGL(KernelName, GridSize, BlockSize, SharedMemorySize, Stream, KernelArguments);". Each of KernelName, GridSize, BlockSize, ShareMemorySize, Stream, and KernelArguments has the same meaning in the HIP kernel startup syntax as it does in the CUDA kernel startup syntax (described earlier in this document). The arguments SharedMemorySize and Stream can be required in the HIP kernel startup syntax, but optional in the CUDA kernel startup syntax.

[0253] A portion of the HIP source code 2430 can be identical to a portion of the CUDA source code 2410 shown, except for the kernel call that causes the kernel MatAdd to execute on the device. The kernel MatAdd can be defined in the HIP source code 2430 using the same "__global__" declaration specifier as used in the CUDA source code 2410. The kernel call in the HIP source code 2430 can be "hipLaunchKernelGGL(MatAdd,numBlocks,threadsPerBlock,0,0,A,B,C);", while the corresponding kernel call in the CUDA source code 2410 is "MatAdd<<<numBlocks,threadsPerBlock> >>(A,B,C);”.

[0254] Other implementations are conceivable, and these can be implemented similarly to the CUDA and HIP implementations described above, such as oneAPI, OpenCL, and other programming platforms. Code can be converted in any direction. For example, CUDA can be converted to HIP, and CUDA can be converted to OpenCL. SnuCL-Tr and CUCL can be used to convert OpenCL to CUDA or CUDA to OpenCL, respectively. Compiled code or intermediate representations (such as CUDA PTX code) can also be converted to run on other processor platforms (such as AMD or Intel). For example, conversion tools (such as ZLUDA) can be used to convert PTX code to run on Intel or AMD processors.

[0255] The techniques described herein can utilize the oneAPI programming model. The oneAPI programming model can refer to a programming model used to interact with various computing accelerator architectures. OneAPI can refer to an application programming interface (API) designed to interact with various computing accelerator architectures. The oneAPI programming model can use the DPC++ programming language. The DPC++ programming language can refer to a high-level language used for data-parallel programming productivity. The DPC++ programming language can be at least partially based on the C and / or C++ programming languages. The oneAPI programming model can be, but is not limited to, a programming model developed by Intel Corporation of Santa Clara, California.

[0256] OneAPI and / or the oneAPI programming model can be used to interact with a variety of accelerators, GPUs, processors, and / or their variant architectures. OneAPI may include a set of libraries that implement various functions. OneAPI may include at least the oneAPI DPC++ library, the oneAPI Math Kernel library, the oneAPI Data Analysis library, the oneAPI Deep Neural Network library, the oneAPI Collection Communication library, the oneAPI Thread Building Block library, the oneAPI Video Processing library, and / or their variants.

[0257] The oneAPI DPC++ library (also known as oneDPL) can be a library that implements algorithms and functions to accelerate DPC++ kernel programming. oneDPL can implement one or more Standard Template Library (STL) functions. OneDPL can implement one or more parallel STL functions. OneDPL can provide a set of library classes and functions, such as, but not limited to, parallel algorithms, iterators, function object classes, range-based APIs, and / or variations thereof. OneDPL can implement one or more classes and / or functions from the C++ Standard Library. OneDPL can implement one or more random number generator functions.

[0258] The oneAPI math kernel library (also known as oneMKL) can be a library that implements various optimized and parallelized routines for a variety of mathematical functions and / or operations. OneMKL can implement one or more Basic Linear Algebra Subroutines (BLAS) and / or Linear Algebra Packages (LAPACK) dense linear algebra routines. OneMKL can implement one or more sparse BLAS linear algebra routines. OneMKL can implement one or more random number generators (RNGs). OneMKL can implement one or more vector math (VM) routines for performing mathematical operations on vectors. OneMKL can implement one or more Fast Fourier Transform (FFT) functions.

[0259] The OneAPI data analysis library (also known as oneDAL) can include libraries for implementing various data analysis applications and distributed computing. OneDAL can implement various algorithms for data analysis preprocessing, transformation, analysis, modeling, validation, and decision-making, including batch, online, and distributed computing processing modes. OneDAL can implement various C++ and / or Java APIs as well as various connectors for connecting to one or more data sources. OneDAL can implement the DPC++ API extension to the traditional C++ interface and support GPU usage for various algorithms.

[0260] The OneAPI Deep Neural Network Library (also known as oneDNN) can include libraries that implement a variety of deep learning functions. OneDNN can implement various neural networks, machine learning and deep learning functions, algorithms and / or variations thereof.

[0261] The OneAPI Collections Communication Library (also known as oneCCL) can include libraries that implement a variety of applications for deep learning and machine learning workloads. OneCCL can be built on lower-level communication middleware, such as, but not limited to, Message Passing Interface (MPI) and libfabrics. OneCCL can support a set of deep learning-specific optimizations, such as, but not limited to, priority ordering, persistence operations, out-of-order execution, and / or variations thereof. OneCCL can implement a variety of CPU and GPU functionalities.

[0262] The OneAPI Thread Building Blocks library (also known as oneTBB) can include libraries that implement various parallelization processes for a wide range of applications. OneTBB can be used for task-based shared parallel programming on a host machine. OneTBB can implement general-purpose parallel algorithms. OneTBB can implement concurrent containers. OneTBB can implement scalable memory allocators. OneTBB can implement work-stealing task schedulers. OneTBB can implement low-level synchronization primitives. OneTBB can be compiler-independent and can be used on a variety of processors, such as, but not limited to, GPUs, PPUs, CPUs, and / or variants thereof.

[0263] The OneAPI video processing library (also known as oneVPL) can include libraries for accelerating video processing in one or more applications. OneVPL can implement a variety of video decoding, encoding, and processing functions. OneVPL can implement various functions for media pipelines on CPUs, GPUs, and other accelerators. OneVPL can implement device discovery and selection in media-centric and video analytics workloads. OneVPL can implement API primitives for zero-copy buffer sharing.

[0264] The oneAPI programming model can use the DPC++ programming language. The DPC++ programming language can include a version of a programming language that is functionally similar to the CUDA mechanism to define device code and distinguish it from host code. The DPC++ programming language can include a subset of the functionality of the CUDA programming language. One or more CUDA programming model operations can be performed using the oneAPI programming model utilizing the DPC++ programming language.

[0265] Any application programming interface (API) described herein can be compiled by a compiler, interpreter, or other software tool into one or more instructions, operations, or other signals. Compilation may include generating one or more machine-executable instructions, operations, or other signals from source code. An API compiled into one or more instructions, operations, or other signals, when executed, may enable one or more processors (e.g., but not limited to...) Figure 9-21B The processor described herein or any other logic circuit further described herein performs one or more computational operations.

[0266] In at least one embodiment, the conversion tools described elsewhere herein (e.g., but not limited to) may include one or more circuitry for converting CUDA code into HIP, oneAPI, OpenCL, or any other language for performing any of the operations described above or elsewhere herein, wherein the CUDA code is used to execute an API to indicate one or more storage locations of one or more scheduled instructions to one or more users; to execute an API to prevent one or more storage locations of one or more scheduled instructions from being stored; and / or to perform operations related to... Figure 1-7 One or more of the processes illustrated or described herein. One or more circuits may be software-configurable to translate CUDA code into HIP, oneAPI, OpenCL, or any other language for performing any of the operations described above or elsewhere herein, the CUDA code being used to execute APIs to cause one or more storage locations of one or more scheduled instructions to be indicated to one or more users; to execute APIs to prevent one or more storage locations of one or more scheduled instructions from being stored; and / or to perform operations related to... Figure 1-7 One or more of the processes illustrated or described.

[0267] Autonomous vehicles Figure 25 An example of an autonomous vehicle 2500 according to at least one embodiment is shown. The autonomous vehicle 2500 (alternately referred to herein as "vehicle 2500") can be a passenger vehicle, such as, but not limited to, a sedan, truck, bus, and / or other type of vehicle capable of accommodating one or more passengers. In at least one embodiment, vehicle 2500 can be a semi-trailer tractor for transporting goods. Vehicle 2500 can be an aircraft, robotic vehicle, or other type of vehicle.

[0268] Autonomous vehicles can be described according to their automation levels, which are defined by the National Highway Traffic Safety Administration (NHTSA) (a division of the U.S. Department of Transportation) and the Society of Automotive Engineers (SAE) in their "Taxonomy and Definitions for Terms Related to Driving Automation Systems for On-Road Motor Vehicles" (e.g., standard number J3016-201806 published June 15, 2018; standard number J3016-201609 published September 30, 2016; and previous and future versions of this standard). In at least one embodiment, vehicle 2500 can achieve functionality at one or more levels of autonomy, from Level 1 to Level 5. For example, in at least one embodiment, depending on the embodiment, vehicle 2500 can achieve conditional automation (Level 3), high automation (Level 4), and / or full automation (Level 5).

[0269] Vehicle 2500 may include components such as, but not limited to, chassis, body, wheels (e.g., 2, 4, 6, 8, 18, etc.), tires, axles, and other vehicle parts. Vehicle 2500 may include a propulsion system 2550, such as, but not limited to, an internal combustion engine, a hybrid power plant, an all-electric motor, and / or other propulsion system types. Propulsion system 2550 may be connected to the drivetrain of vehicle 2500 (which may include a transmission) to propel vehicle 2500. Propulsion system 2550 may be controlled based on signals received from throttle / accelerator 2552.

[0270] A steering system 2554 (which may include a steering wheel) is used to maneuver the vehicle 2500 (e.g., to travel along a desired path or route) while the propulsion system 2550 is in operation (e.g., when the vehicle 2500 is in motion). The steering system 2554 may receive signals from the steering actuator 2556. The steering wheel is optional for fully automated (Level 5) functionality. A brake sensor system 2546 may be used to operate the vehicle brakes in response to signals received from the brake actuator 2548 and / or brake sensors.

[0271] Controller 2536 may include one or more system-on-a-chip (“SoC”) and / or graphics processing units (“GPUs”) that can provide signals (e.g., signals representing commands) to one or more components and / or systems of vehicle 2500. For example, controller 2536 may send signals to operate vehicle brakes via brake actuator 2548, steering system 2554 via steering actuator 2556, and propulsion system 2550 via throttle / accelerator 2552. Controller 2536 may include one or more onboard (e.g., integrated) computing devices for processing sensor signals and outputting operational commands (e.g., signals representing commands) to enable autonomous driving and / or assist a human driver in driving vehicle 2500. Controller 2536 may include a first controller for autonomous driving functions, a second controller for functional safety functions, a third controller for artificial intelligence functions (e.g., computer vision), a fourth controller for infotainment functions, a fifth controller for redundancy in emergency situations, and / or other controllers. A single controller may handle two or more of the above functions, two or more controllers may handle a single function, and / or any combination thereof.

[0272] The controller 2536 may provide signals for controlling one or more components and / or systems of the vehicle 2500 in response to sensor data (e.g., sensor inputs) received from one or more sensors. Sensor data can be received from, for example, a Global Navigation Satellite System (“GNSS”) sensor 2558 (e.g., a Global Positioning System sensor), a RADAR (radar) sensor 2560, an ultrasonic sensor 2562, a LIDAR (light radar) sensor 2564, an Inertial Measurement Unit (“IMU”) sensor 2566 (e.g., an accelerometer, a gyroscope, one or more magnetic compasses, a magnetometer, etc.), a microphone 2596, a stereo camera 2568, a wide-angle camera 2570 (e.g., a fisheye camera), an infrared camera 2572, a surround-view camera 2574 (e.g., a 360-degree camera), a long-range camera 2598, a medium-range camera 2576, a speed sensor 2544 (e.g., for measuring the speed of vehicle 2500), a vibration sensor 2542, a steering sensor 2540, a braking sensor (e.g., as part of a braking sensor system 2546), and / or other types of sensors.

[0273] One or more controllers 2536 may receive input (e.g., represented by input data) from the instrument panel 2532 of the vehicle 2500 and provide output (e.g., represented by output data, display data, etc.) via a human-machine interface (“HMI”) display 2534, an audible annunciator, a speaker, and / or via other components of the vehicle 2500. Output may include, but is not limited to, vehicle speed, rate, time, map data (e.g., a high-resolution map (not shown), location data (e.g., the location of the vehicle 2500, such as, but not limited to, its location on a map), direction, the location of other vehicles (e.g., occupying a grid), information about objects, and the state of objects perceived by the controllers 2536. For example, the HMI display 2534 may display information about the presence of one or more objects (e.g., street signs, warning signs, traffic light changes, etc.) and / or information about driving actions that the vehicle has performed, is performing, or will perform (e.g., changing lanes now, exiting from exit 34B in two miles, etc.).

[0274] Figure 25 Every component, feature, and system of the vehicle 2500 can be connected via bus 2502. Bus 2502 may include a CAN data interface (also referred to herein as the "CAN bus"). CAN can be a network within the vehicle 2500 used to assist in the control of various features and functions of the vehicle 2500, such as, but not limited to, the activation of brakes, acceleration, braking, steering, windshield wipers, etc. Bus 2502 can be configured to have dozens or even hundreds of nodes, each with its own unique identifier (e.g., a CAN ID). Bus 2502 can be read to locate steering wheel angle, ground speed, engine revolutions per minute ("RPM"), button positions, and / or other vehicle status indicators. Bus 2502 can be an ASIL B compliant CAN bus.

[0275] In addition to CAN, or as an alternative to CAN, FlexRay and / or Ethernet protocols may be used. Bus 2502 may consist of any number of buses, which may include zero or more CAN buses, zero or more FlexRay buses, zero or more Ethernet buses, and / or zero or more other types of buses using different protocols. Two or more buses may be used to perform different functions and / or for redundancy. For example, a first bus may be used for collision avoidance functions, and a second bus may be used for actuation control. Each bus in bus 2502 may communicate with any component of vehicle 2500, and two or more buses of bus 2502 may communicate with corresponding components. Any number of System-on-Chip (“SoC”) 2504 (e.g., but not limited to SoC 2504(A) and SoC 2504(B)), each controller 2536, and / or each computer within the vehicle may access the same input data (e.g., input from sensors of vehicle 2500) and may be connected to a common bus, such as a CAN bus.

[0276] According to at least one embodiment, for Figure 25 The autonomous vehicle 2500 can place any number of cameras at any chosen camera locations and within any field of view. The cameras and corresponding fields of view are an exemplary embodiment and are not intended to be limiting. For example, additional and / or alternative cameras may be included, and / or the cameras may be located at different locations within the vehicle 2500.

[0277] The camera type may include a digital camera suitable for components and / or systems in vehicle 2500. The camera may operate under Automotive Safety Integrity Level (“ASIL”) B and / or other ASILs. According to embodiments, the camera type may support any image capture rate, such as, but not limited to, 60 frames per second (fps), 1220 fps, 240 fps, etc. The camera may be able to use a rolling shutter, a global shutter, other types of shutters, or combinations thereof. In at least one embodiment, the color filter array may include a red-to-clear-to-clear (“RCCC”) color filter array, a red-to-clear-to-blue (“RCCB”) color filter array, a red-blue-green (“RBGC”) color filter array, a Foveon X3 color filter array, a Bayer sensor (“RGGB”) color filter array, a monochrome sensor color filter array, and / or other types of color filter arrays. Transparent pixel cameras (e.g., but not limited to cameras with RCCC, RCCB, and / or RBGC color filter arrays) may be used to enhance sensitivity.

[0278] One or more cameras may be used to perform advanced driver assistance system (“ADAS”) functions (e.g., as part of a redundancy or fail-safe design). For example, a multi-function monochrome camera may be installed to provide functions including lane departure warning, traffic sign assist, and intelligent headlight control. One or more cameras (e.g., all cameras) may simultaneously record and provide image data (e.g., video).

[0279] One or more cameras can be mounted in mounting assemblies, such as, but not limited to, custom-designed (3D-printed) assemblies, to eliminate stray light and reflections inside the vehicle 2500 (e.g., dashboard reflections in the windshield mirror) that could interfere with the camera's image data capture capabilities. Regarding rearview mirror mounting assemblies, these assemblies can be custom-3D printed to match the shape of the camera mounting plate to the rearview mirror. The camera can be integrated into the rearview mirror. For side-view cameras, the cameras can also be integrated into the four pillars in each corner of the cab.

[0280] The field of view includes a portion of the environment in front of the vehicle 2500 that can be viewed by cameras (e.g., a front-facing camera) to help identify the path and obstacles ahead, and, with the assistance of one or more controllers 2536 and / or control SoCs, to provide information crucial for generating an occupancy grid and / or determining a preferred vehicle path. The front-facing camera can be used to perform many ADAS functions similar to LiDAR, including emergency braking, pedestrian detection, and collision avoidance. The front-facing camera can also be used in ADAS functions and systems, including lane departure warning (“LDW”), adaptive cruise control (“ACC”), and / or other functions such as, but not limited to, traffic sign recognition.

[0281] Various front-mounted cameras can be used, including, for example, monocular camera platforms including CMOS (“complementary metal-oxide-semiconductor” color imagers. A wide-angle camera 2570 can be used to perceive objects entering the field of view from the periphery (e.g., pedestrians, crossing vehicles, or bicycles). Any number (including zero) of wide-angle cameras 2570 can be mounted on the vehicle 2500. Any number of remote cameras 2598 (e.g., long-view stereo camera pairs) can be used for depth-based object detection, particularly for objects on which neural networks have not yet been trained. The remote cameras 2598 can also be used for object detection and classification, as well as basic object tracking.

[0282] It may also include any number of front-mounted stereo cameras 2568. One or more stereo cameras 2568 may include an integrated control unit that includes a scalable processing unit that can provide programmable logic (“FPGA”) and a multi-core microprocessor with an integrated controller area network (“CAN”) or Ethernet interface on a single chip. Such a unit can be used to generate a 3D map of the environment of the vehicle 2500, including distance estimates for all points in the image. One or more stereo cameras 2568 may include a compact stereo vision sensor that may include two camera lenses (one on each side) and an image processing chip that measures the distance from the vehicle 2500 to a target object and uses the generated information (e.g., metadata) to activate autonomous emergency braking and lane departure warning functions. Other types of stereo cameras 2568 may be used in addition to, or as an alternative to, the stereo cameras described herein.

[0283] Cameras (e.g., side-view cameras) that include portions of the environment along the sides of vehicle 2500 can be used in the surround view to provide information for creating and updating occupancy grids and generating side collision warnings. For example, surround view cameras 2574 (e.g., four surround view cameras) can be mounted on vehicle 2500. Surround view cameras 2574 can include any number and combination of wide-angle cameras, fisheye cameras, 360-degree cameras, and / or similar cameras. For example, four fisheye cameras can be mounted at the front, rear, and sides of vehicle 2500, respectively. Vehicle 2500 can use three surround view cameras 2574 (e.g., left, right, and rear) and can utilize one or more other cameras (e.g., front-facing cameras) as a fourth surround view camera.

[0284] A field of view including a portion of the environment behind vehicle 2500 (e.g., a rear-view camera) can be used for parking assistance, surround view, rear collision warning, and creating and updating an occupancy grid. A wide variety of cameras can be used, including but not limited to cameras that are also suitable as front-facing cameras (e.g., long-range camera 2598 and / or mid-range camera 2576, stereo camera 2568, infrared camera 2572, etc.), as described herein.

[0285] Vehicle 2500 may include any number of SoC 2504 or other processors described elsewhere in this document, such as, but not limited to, those described elsewhere. Figure 9-21BThe processors and / or components shown and described. Each SoC 2504 may include a central processing unit (“CPU”) 2506, a graphics processing unit (“GPU”) 2508, a processor 2510, a cache 2512, an accelerator 2514, a data storage 2516, and / or other components and features not shown. SoC 2504 can be used to control vehicle 2500 in various platforms and systems. For example, SoC 2504 may be combined with a high-definition (“HD”) map 2522 in a system (e.g., the system of vehicle 2500), which may obtain map refreshes and / or updates from one or more servers (not shown) via network interface 2524. SoC 2504 may include logic 2515, which may include any combination of software logic, hardware logic, and / or firmware logic for providing the functions or operations described herein, wherein the logic may be collectively or separately embodied as circuitry forming part of a larger system, such as an integrated circuit (IC), a system-on-a-chip (SoC), or one or more processors (e.g., CPU, GPU).

[0286] The CPU 2506 may include a CPU cluster or CPU complex (referred to herein as “CCPLEX”). The CPU 2506 may include multiple cores and / or a secondary (“L2”) cache. For example, the CPU 2506 may include eight cores in a consistent multiprocessor configuration. The CPU 2506 may include four dual-core clusters, each with a dedicated L2 cache (e.g., 2 megabytes (MB) of L2 cache). The CPU 2506 (e.g., CCPLEX) may be configured to support simultaneous cluster operation, allowing any combination of CPU 2506 clusters to be active at any given time.

[0287] One or more of the CPU 2506 can implement power management capabilities, including one or more of the following features: individual hardware blocks can automatically clock-gated when idle to save dynamic power; clock gating can be applied to cores when each core is not actively executing instructions due to executing Wait for Interrupt (“WFI”) / Wait for Event (“WFE”) instructions; each core can independently perform power gating; each core cluster can independently clock-gated when all cores can perform clock or power gating; and / or each core cluster can independently power-gated when all cores can perform power gating. The CPU 2506 can also implement enhanced algorithms for managing power states, where allowed power states and expected wake-up times can be specified, and the hardware / microcode determines which optimal power state the core, cluster, and CCPLEX enter. Processing cores can enter a simplified power state sequence via software support, offloading work to the microcode.

[0288] The GPU 2508 may include an integrated GPU (referred to herein as an "iGPU"). The GPU 2508 may be programmable and capable of efficiently handling parallel workloads. The GPU 2508 may use an enhanced tensor instruction set. The GPU 2508 may include one or more streaming microprocessors, each of which may include a Level 1 ("L1") cache (e.g., an L1 cache with a storage capacity of at least 96KB), and two or more streaming microprocessors may share an L2 cache (e.g., an L2 cache with a storage capacity of 512KB). The GPU 2508 may include at least eight streaming microprocessors. The GPU 2508 may use a computation application programming interface (API). The GPU 2508 may use one or more parallel computing platforms and / or programming models (e.g., NVIDIA's CUDA model). A streaming microprocessor may be referred to as a streaming multiprocessor (“SM”), a streaming processor (“SP”), a streaming processing unit (“SPU”), a compute unit (“CU”), an execution unit (“EU”), and / or a slice, wherein, in this context, a slice may refer to a portion of the processing resources within a processing unit (e.g., 16 cores, a ray tracing unit, a thread bootstrap, or a scheduler).

[0289] One or more of the GPU 2508 can be power-optimized for optimal performance in automotive and embedded use cases. For example, the GPU 2508 can be fabricated on FinFET (“FinFET”) circuitry. Each streaming microprocessor can include multiple mixed-precision processing cores partitioned into blocks. For example, 64 PF32 cores and 32 FP64 cores can be partitioned into four processing blocks. Each processing block can be allocated 16 FP32 cores, 8 FP64 cores, 16 INT32 cores, two mixed-precision NVIDIA Tensor cores for deep learning matrix operations, a level-zero (“L0”) instruction cache, a scheduler (e.g., a warp scheduler) or sequencer, dispatch units, and / or a 64KB register file. Streaming microprocessors can include independent parallel integer and floating-point data paths for efficiently performing workloads involving mixed computation and addressing computation. Streaming microprocessors can include independent thread scheduling capabilities to enable finer-grained synchronization and cooperation between parallel threads. Streaming microprocessors may include a combination of L1 data cache and shared memory units to improve performance while simplifying programming.

[0290] One or more GPU 2508s may include high-bandwidth memory (“HBM”) and / or a 16GB HBM2 memory subsystem, used in some examples to provide a peak memory bandwidth of approximately 900GB / s. In addition to HBM memory, or as an alternative to HBM memory, synchronous graphics random access memory (“SGRAM”) may be used, such as, but not limited to, graphics double data rate type 5 synchronous random access memory (“GDDR5”).

[0291] The GPU 2508 may include unified memory technology. Address translation service (“ATS”) supports page tables that allow the GPU 2508 to directly access the CPU 2506. When a GPU memory management unit (“MMU”) miss occurs in one or more GPUs 2508, an address translation request can be sent to the CPU 2506. In response, two of the CPUs 2506 can look up the virtual-to-physical mapping of the address in their page tables and transfer the translation back to the GPU 2508. Unified memory technology can provide a single, unified virtual address space for the memory of both the CPU 2506 and GPU 2508, thereby simplifying GPU 2508 programming and application porting to the GPU 2508.

[0292] The GPU 2508 may include any number of access counters that can track the frequency of the GPU 2508's access to the memory of other processors. These access counters help ensure that memory pages are moved to the physical memory of the processor that accesses those pages most frequently, thereby improving the efficiency of shared memory ranges between processors.

[0293] One or more SoCs 2504 may include any number of caches 2512, including the caches described herein. For example, cache 2512 may include a Level 3 (“L3”) cache available for use by both CPU 2506 and GPU 2508 (e.g., connected to both CPU 2506 and GPU 2508). Cache 2512 may include a write-back cache that can track the state of rows, for example, but not limited to, by using a cache coherence protocol (e.g., MEI, MESI, MSI, etc.). According to embodiments, the L3 cache may include 4 MB or more of memory, although a smaller cache size may be used.

[0294] One or more SoCs 2504 may include one or more accelerators 2514 (e.g., hardware accelerators, software accelerators, or a combination thereof). SoC 2504 may include a hardware acceleration cluster that may include optimized hardware accelerators and / or large on-chip memory. Large on-chip memory (e.g., 4MB SRAM) enables the hardware acceleration cluster to accelerate neural networks and other computations. The hardware acceleration cluster can be used to supplement the GPU 2508 and offload some tasks from the GPU 2508 (e.g., freeing up more cycles of the GPU 2508 to perform other tasks). Accelerators 2514 can be used for targeted workloads (e.g., perceptrons, convolutional neural networks (“CNNs”), recurrent neural networks (“RNNs”), etc.) that are sufficiently stable to be easily accelerated. CNNs may include region-based or regional convolutional neural networks (“RCNNs”) and fast RCNNs (e.g., for object detection) or other types of CNNs.

[0295] Accelerator 2514 (e.g., a hardware acceleration cluster) may include one or more deep learning accelerators (“DLAs”). A DLA may include one or more tensor processing units (“TPUs”) that can be configured to provide trillions of operations per second for deep learning applications and inference, such as those described in this paper (e.g., Figure 19 The TPU described in the text is a TPU that can be configured and optimized to perform image processing functions, such as CNNs, RCNNs, etc. The DLA can be further optimized for a specific set of neural network types, floating-point operations, and inference. The DLA is designed to provide higher performance per millimeter than typical general-purpose GPUs and often significantly outperforms CPUs. The TPU can perform several functions, including single-instance convolutional functions, support for features and weights of data types such as INT8, INT16, and FP16, and post-processor functions. The DLA can quickly and efficiently execute neural networks (especially CNNs) on processed or unprocessed data to achieve a variety of functions, including, for example: CNNs for object recognition and detection using data from camera sensors; CNNs for distance estimation using data from camera sensors; CNNs for emergency vehicle detection and recognition using data from microphones; CNNs for face recognition and driver identification using data from camera sensors; and / or CNNs for safety and / or safety-related events.

[0296] The DLA can perform any function of the GPU 2508, and by using the inference accelerator, designers can direct any function to either the DLA or the GPU 2508. For example, designers can focus CNN and floating-point processing on the DLA, while leaving other functions to the GPU 2508 and / or the accelerator 2514.

[0297] Accelerator 2514 may include a programmable vision accelerator (“PVA”), which may alternatively be referred to herein as a computer vision accelerator. The PVA may be designed and configured to accelerate computer vision algorithms for advanced driver assistance systems (“ADAS”) 2538, autonomous driving, augmented reality (“AR”) applications, and / or virtual reality (“VR”) applications. The PVA can provide a balance between performance and flexibility. For example, each PVA may include any number of reduced instruction set computer (“RISC”) cores, direct memory access (“DMA”), and / or any number of vector processors.

[0298] The RISC core can interact with image sensors (e.g., the image sensor of any camera described herein), image signal processors, etc. Each RISC core may include memory of any size. The RISC core can use any of a variety of protocols, depending on the implementation. The RISC core can execute a real-time operating system (“RTOS”). The RISC core can be implemented using one or more integrated circuit devices, application-specific integrated circuits (“ASICs”), and / or memory devices. For example, the RISC core may include an instruction cache and / or tightly coupled RAM.

[0299] DMA enables PVA components to access system memory independently of the CPU 2506. DMA supports any number of features for providing optimizations to the PVA, including support for multidimensional addressing and / or circular addressing. DMA supports addressing in up to six or more dimensions, which can include block width, block height, block depth, horizontal block step, vertical block step, and / or depth step.

[0300] The vector processor can be a programmable processor designed to efficiently and flexibly execute computer vision algorithms and provide signal processing capabilities. The PVA may include a PVA core and two vector processing subsystem partitions. The PVA core may include a processor subsystem, a DMA engine (e.g., two DMA engines), and / or other peripherals. The vector processing subsystem may operate as the main processing engine of the PVA and may include a vector processing unit (“VPU”), an instruction cache, and / or a vector memory (e.g., “VMEM”). The VPU core may include a digital signal processor, such as, but not limited to, a Single Instruction Multiple Data (“SIMD”) or Very Long Instruction Word (“VLIW”) digital signal processor. The combination of SIMD and VLIW can improve throughput and speed.

[0301] Each vector processor may include an instruction cache and may be coupled to dedicated memory. Therefore, each vector processor can be configured to execute independently of other vector processors. The vector processors included in a particular PVA can be configured to employ data parallelism. For example, multiple vector processors included in a single PVA can execute general computer vision algorithms, but for different regions of an image. The vector processors included in a particular PVA can execute different computer vision algorithms on a single image simultaneously, or even execute different algorithms on consecutive images or multiple parts of an image. Furthermore, a hardware-accelerated cluster can include any number of PVAs, and each PVA can include any number of vector processors. PVAs may include additional error-correcting code (“ECC”) memory to enhance overall system security.

[0302] Accelerator 2514 may include an on-chip computer vision network and static random access memory (“SRAM”) to provide high-bandwidth, low-latency SRAM for accelerator 2514. The on-chip memory may include at least 4 MB of SRAM, for example, comprising eight field-configurable mem...

Claims

1. A processor, comprising: One or more circuits, the one or more circuits being configured to execute an application programming interface (API) to indicate one or more storage locations of one or more scheduled instructions to one or more users.

2. The processor of claim 1, wherein the one or more storage locations include one or more addresses of the one or more scheduled instructions in one or more buffers.

3. The processor of claim 1, wherein the one or more scheduled instructions correspond to one or more cores to be executed by one or more graphics processing units (GPUs).

4. The processor of claim 1, wherein the one or more circuits are configured to indicate the one or more storage locations in the one or more user-accessible data structures.

5. The processor of claim 1, wherein the one or more circuits are configured to indicate one or more semaphore values ​​associated with the one or more scheduled instructions to the one or more users.

6. The processor of claim 1, wherein the one or more circuits are further configured to: Obtain instructions on which kernel will be executed by one or more graphics processing units (GPUs); and This ensures that at least one or more addresses used to execute one or more instructions of the kernel are indicated to the one or more users.

7. The processor of claim 1, wherein the one or more scheduled instructions are scheduled to be executed by one or more device processing units.

8. A method comprising: The instruction received by the application programming interface (API) causes one or more storage locations of one or more scheduled instructions to be indicated to one or more users; The instructions for the one or more scheduled instructions are obtained at least in part based on computer program code; as well as This causes the one or more storage locations of the one or more scheduled instructions to be stored in one or more data structures.

9. The method of claim 8, further comprising: Obtain instructions for one or more kernels to be executed by one or more graphics processing units (GPUs); as well as This causes one or more addresses for executing one or more methods of the one or more kernels to be indicated to the one or more users.

10. The method of claim 8, further comprising: This causes one or more channels associated with the one or more scheduled instructions to be directed to the one or more users.

11. The method of claim 8, wherein the one or more storage locations include one or more start addresses and end addresses of one or more buffers.

12. The method of claim 8, wherein one or more input parameters of the API include an indication of an operation flow to be performed by one or more processing units.

13. The method of claim 8, further comprising: This causes one or more semaphore values ​​associated with the execution of the one or more scheduled instructions to be indicated to the one or more users.

14. A system comprising: One or more processors, the one or more processors being configured to execute an application programming interface (API) to indicate one or more storage locations of one or more scheduled instructions to one or more users.

15. The system of claim 14, wherein the one or more scheduled instructions are associated with one or more kernels scheduled to execute after the API is invoked.

16. The system of claim 14, wherein the one or more processors are configured to use a tracking data structure to indicate the one or more storage locations to the one or more users.

17. The system of claim 14, wherein the API causes the one or more processors to begin indicating the storage location of subsequently scheduled instructions to the one or more users.

18. The system of claim 14, wherein the one or more scheduled instructions are scheduled by the host processing unit for execution by the device processing unit.

19. The system of claim 14, wherein the one or more storage locations include indications of semaphores associated with the one or more scheduled instructions.

20. The system of claim 14, wherein the return of the API includes one or more status indications.

Images