Control device, control method and distribution processing system
The control device addresses load imbalances in FPGAs by using pre-synthesized templates to reconfigure circuit logic, improving processing efficiency and reducing delays in domain-oriented systems.
Patent Information
- Application Number
- JP2023071252
- Authority / Receiving Office
- JP · JP
- Patent Type
- Applications
- Current Assignee / Owner
- Filing Date
- 2023-04-25
- Publication Date
- 2024-11-07
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
In domain-oriented systems using FPGAs, load imbalances among processes lead to inefficient processing due to the time-consuming nature of rewriting circuit logic, causing delays and reduced performance.
A control device with an optimization unit that rapidly reconfigures FPGAs by reading pre-synthesized circuit templates from a storage area and writing them into appropriate circuit areas to balance load, utilizing a logical pool to manage and distribute processing efficiently.
This approach enhances processing efficiency by dynamically balancing load among FPGAs, reducing delays and optimizing performance in real-time without the lengthy synthesis and placement processes.
Smart Images

Figure 2024157115000001_ABST
Abstract
Description
[Technical field]
[0001] The present invention relates to a control device, a control method, and a distributed processing system. [Background technology]
[0002] 2. Description of the Related Art In recent years, the improvement in performance of semiconductors such as CPUs (Central Processing Units) has slowed, and the improvement in computer performance due to the high performance of semiconductors is approaching its limit.
[0003] On the other hand, with the development of AI (Artificial Intelligence) technology and IoT (Internet of Things), the amount of data to be processed is increasing, and applications run by computers are expected to perform tasks at faster speeds and with greater volumes of data than ever before.
[0004] One of the methods to speed up applications is a system that uses a domain-oriented architecture (hereinafter referred to as a "domain-oriented system"). Domain-oriented architecture is an approach that improves the performance and operability of a server by narrowing down the application domain to which it is applied and optimizing the hardware and software according to the characteristics of that domain.
[0005] In domain-oriented systems, ASICs (Application Specific Integrated Circuits), which are accelerators specialized for specific calculations in specific fields, are used. Since ASICs are manufactured as dedicated hardware, their usage is limited to specific applications. This means that it is difficult to apply domain-oriented systems from the perspective of design and manufacturing costs unless the application is commercially viable.
[0006] In order to facilitate the application of domain-oriented systems to applications, FPGAs (Field-Programmable Gate Arrays) have been attracting attention as reprogrammable ASICs. FPGAs are an example of PLDs (Programmable Logic Devices). PLDs may also be called programmable logic circuits.
[0007] FPGAs have a circuit area where circuit logic is written, and the reconfigurable function allows the circuit logic in the circuit area to be rewritten. Circuit logic is information that indicates the structure of a logical circuit that enables the FPGA to perform a specific process. Since various types of ASICs can be created by rewriting the circuit logic in the circuit area, using FPGAs in a domain-oriented system makes it possible to balance design and manufacturing costs with performance. [Prior art documents] [Patent documents]
[0008] [Patent Document 1] US Patent Application Publication No. 2019 / 0334771 [Patent Document 2] US Patent Application Publication No. 2020 / 0099579 [Patent Document 3] JP 2015-184549 A [Patent Document 4] JP 2018-132981 A Summary of the Invention [Problem to be solved by the invention]
[0009] For example, in a domain-oriented system (distributed processing system) in which multiple processes are distributed and executed by one or more FPGAs, consider a case in which the load between multiple processes is unbalanced, such as when the number of execution requests for process A is large and the number of execution requests for process B is small.
[0010] In such a case, while the FPGA in which the circuit logic of process B is written has ample processing resources, the FPGA in which the circuit logic of process A is written may have insufficient processing resources, causing process A to stagnate (wait for processing). This decrease in the processing efficiency (operational efficiency) of the FPGA may cause processing delays in the computer that requested the execution of process A.
[0011] Therefore, it is conceivable that the load imbalance can be resolved by using the reconfigurable function to write the circuit logic of process A into the circuit area of one or more FPGAs. However, rewriting the circuit logic of an FPGA takes time. For example, the synthesis and place and route steps of the rewriting process can take several days. For this reason, it is difficult to change the processes to be executed by the FPGA in real time, and the load imbalance may not be resolved.
[0012] In this way, the processing efficiency of the FPGA may decrease depending on the load of the processing executed by the FPGA, which may cause processing delays in the computer that originated the execution request.
[0013] In one aspect, an object of the present invention is to improve the processing efficiency of a programmable logic circuit. [Means for solving the problem]
[0014] In one aspect, the control device may include a control unit that performs the following processes. When the control unit detects a load imbalance between a plurality of processes executed by circuits set in rewritable circuit areas included in each of one or more programmable logic circuits, the control unit may read a first template that satisfies a preset condition from a storage area that stores a plurality of circuit logic templates for which synthesis of a circuit design has been completed and placement and wiring has been determined. The control unit may also write the read first template into a first circuit area that satisfies the condition among the one or more circuit areas, thereby setting a first process that is realized by the circuit logic of the first template in the first circuit area. Effect of the Invention
[0015] In one aspect, the present invention can improve the processing efficiency of a programmable logic circuit. [Brief description of the drawings]
[0016] [Figure 1] FIG. 1 is a block diagram showing a configuration of a system according to a comparative example. [Diagram 2] FIG. 13 is a diagram for explaining an example of a process for writing circuit logic to an FPGA. [Diagram 3] 1 is a block diagram showing an example of the configuration of a system according to an embodiment; [Figure 4] FIG. 2 is a block diagram showing an example of a hardware configuration of a system according to an embodiment. [Diagram 5] FIG. 2 is a block diagram showing an example of a software configuration of an optimization device according to an embodiment. [Figure 6] FIG. 11 is a diagram illustrating an example of determination condition management information. [Figure 7] 11 is a diagram for explaining an example of a determination process performed by a determination unit; FIG. [Figure 8] FIG. 11 is a diagram illustrating an example of FPGA management information. [Figure 9] FIG. 13 illustrates an example of template management information. [Figure 10]FIG. 11 is a sequence diagram illustrating an example of the operation of a system according to an embodiment. [Figure 11] FIG. 13 illustrates an example of updating FPGA management information. [Figure 12] FIG. 13 illustrates an example of updating template management information. [Figure 13] FIG. 13 illustrates an example of updating judgment condition management information. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS
[0017] Hereinafter, an embodiment of the present invention will be described with reference to the drawings. However, the embodiment described below is merely an example, and is not intended to exclude various modifications or application of techniques not specified below. For example, this embodiment can be modified in various ways without departing from the spirit of the invention. In the drawings used in the following description, parts with the same reference numerals represent the same or similar parts unless otherwise specified.
[0018] [A] Comparative example Fig. 1 is a block diagram showing the configuration of a system 100 according to a comparative example. System 100 illustratively includes a plurality of cameras 110 (cameras #0 to #n-1) (n cameras in the example of Fig. 1: n is an integer of 2 or more), a host PC (Personal Computer) 120, an interface 130, and one or more FPGAs 140 (FPGA group) (four in the example of Fig. 1).
[0019] The host PC 120 executes AI processing used in autonomous driving or security systems, etc., using captured images captured by the multiple cameras 110. The AI processing may be, for example, image recognition processing such as object detection processing. For the AI processing, images that have been subjected to pre-processing (image processing) such as edge extraction processing and binarization processing, rather than the captured image itself, are input.
[0020] The pre-processing is a process that places a high execution load on the host PC 120. For this reason, a processing delay occurs when the host PC 120 executes the pre-processing. Therefore, in the system 100 according to the comparative example, the execution of the pre-processing is offloaded from the host PC 120 to the FPGA 140.
[0021] For example, the host PC 120 outputs a captured image to the FPGA 140 via a high-speed communication path connected to the interface 130 or an IO (Input / Output) port (not shown) of the FPGA 140. Then, the host PC 120 obtains the result of pre-processing executed at high speed by the FPGA 140 and performs AI processing using the result.
[0022] The FPGA 140 has a circuit area where the circuit logic can be rewritten, and executes a predetermined process by a circuit set in the circuit area. Hereinafter, when distinguishing between the FPGAs 140, they will be referred to as FPGA #0 to #3 (see FIG. 1). For example, each of the FPGAs #0 to #3 may execute the following process.
[0023] FPGA#0: Processing A: Edge extraction processing for 4K resolution captured images FPGA#1: Processing B: Edge extraction processing for HR (High Resolution) captured images FPGA#2: Processing C: Edge extraction processing for captured images with STD (Standard) resolution FPGA#3: Processing D: Binarization
[0024] In the example of Fig. 1, captured images are input to FPGAs #0 to #2 according to resolution. Edges extracted from the captured images by each of FPGAs #0 to #2 (processing A to C) are input to FPGA #3 and binarized by FPGA #3 (processing D). The binarized image data is output to host PC 120 as the processing result of the pre-processing.
[0025] Processes A to C are all edge extraction processes for captured images, but they differ in the resolution (maximum resolution) that they can handle. In the example of Fig. 1, system 100 causes FPGAs #0 to #2 to execute the same edge extraction processes for each resolution, which can easily lead to a redundant configuration.
[0026] The scale of the circuit set in the rewritable circuit area of the FPGA 140 differs depending on the contents of the processes A to D executed by the circuit. The circuit scale of each of the processes A to D is indicated by the size of the frame at reference numeral 141 in FIG.
[0027] For example, the higher the resolution is, the larger the circuit scale 141 becomes in the edge extraction processes of processes A to C. In the example of Fig. 1, the order of resolution that can be handled is process A (4K), process B (HR), and process C (STD), and the order of circuit scale 141 is process A, process B, and process C. Note that process D is a process that uses the processing results output sequentially from each of processes A to C, and since it receives edges based on captured images with a maximum resolution of 4K, it shows a circuit scale 141 that matches the circuit scale of process A.
[0028] For this reason, for example, when changing the circuit of each FPGA 140 using a reconfigurable function, it is important to determine an appropriate combination of processes based on the size of the free space in the rewritable circuit area of each FPGA 140 and the circuit scale 141 of each process.
[0029] Next, a description will be given of a process for rewriting (writing) the circuit logic to the FPGA 140, that is, a process for setting the circuit.
[0030] As shown in Fig. 2, a developer inputs a circuit logic design in a language such as HDL (Hardware Description Language) into a development environment (step P1), and performs design synthesis using a logic synthesis tool (step P2). At this time, the developer performs design verification by performing logic verification of the design (step P3), and completes design synthesis.
[0031] After completing the design synthesis, the developer performs design implementation to incorporate a logic circuit on the FPGA 140 (step P4).
[0032] For example, the developer optimizes each function to be implemented on the rewritable circuit area (step P41), and places and routes each optimized function on the circuit area (step P42). At this time, the developer performs static timing analysis (step P5) and timing verification (step P7) using back annotation (step P6) to determine the placement and routing.
[0033] Then, based on the determined placement and wiring, the developer creates a bit stream for writing (programming) the circuit in the circuit area (step P43).
[0034] The developer downloads the bitstream to the FPGA 140 (step P8). For example, the developer writes the bitstream to an external storage device of the FPGA 140 via an interface conforming to a standard such as JTAG (Joint Test Action Group). Step P8 may also be referred to as configuring the FPGA 140, writing a logic circuit, or the like.
[0035] The bit stream written to the external storage device is read from the external storage device when the FPGA 140 is started up or when the reconfigurable function is executed, and is written to a rewritable circuit area of the FPGA 140. After writing the bit stream to the circuit area, the developer performs actual device verification (step P9). This completes the circuit setting in the FPGA 140.
[0036] For example, if a load imbalance occurs between processes A to D executed by the FPGA 140, the operating efficiency of the FPGA 140 will deteriorate, which may cause a delay in the execution of the AI processing by the host PC 120. In order to resolve such an imbalance, for example, a circuit that executes a high-load process may be set in a circuit area of the FPGA 140 in which a circuit that executes a low-load process is set, thereby attempting to resolve the load imbalance.
[0037] However, as described above, the synthesis, placement, and wiring steps (e.g., steps P2 and P42 in FIG. 2) in the rewrite process may take several days. For this reason, it is difficult to change the process to be executed by the FPGA 140 in real time.
[0038] Therefore, in one embodiment, an example of a method for improving the processing efficiency of an FPGA will be described.
[0039] [B] Description of a system according to one embodiment Fig. 3 is a block diagram showing an example of the configuration of a system 1 according to an embodiment. The system 1 is an example of a distributed processing system or an information processing system. The system 1 illustratively includes a plurality of cameras 2 (cameras #0 to #n-1) (n cameras in the example of Fig. 3), a host PC 3, an optimization device 4, an interface 6, and one or more FPGAs 7 (FPGA group) (four cameras in the example of Fig. 3).
[0040] The system 1 according to one embodiment causes the FPGA 7 to execute a plurality of processes that are pre-processing of the AI processing. The pre-processing may include, for example, edge extraction processing for an image, binarization processing based on the extracted edges, etc. Note that the plurality of processes executed by the FPGA 7 are not limited to the pre-processing of the AI processing, and may include various types of processing.
[0041] The camera 2 is an example of an imaging device that captures an image of a predetermined imaging area and outputs the captured image. Each of the multiple cameras 2 may output captured images with different resolutions.
[0042] The host PC 3 is an example of an information processing device or a computer, and includes an AI processing engine 30 that performs AI processing on an image captured by the camera 2. The host PC 3 offloads pre-processing to the FPGA 7, for example, and performs AI processing using the AI processing engine 30 based on the results of the pre-processing by the FPGA 7.
[0043] For example, when the host PC 3 acquires a captured image from the camera 2, it may transmit a request for execution of pre-processing (processing request) including the acquired captured image. The execution request may be transmitted in a form that can be acquired by the optimization device 4. The AI processing may be, for example, an image recognition processing such as object detection processing, and may be used in an automatic driving or security system, etc.
[0044] The host PC 3 and the optimization device 4 may be connected to the IO port of the FPGA 7 via a communication path 1a so that they can communicate with each other. The communication path 1a may be used for processing such as transmitting a pre-processing execution request and the execution result, and writing the communication path 1a to the circuit area 70. For example, the host PC 3 may send an execution request to the optimization device 4 via a network or the communication path 1a. The host PC 3 may receive the pre-processing execution result from the FPGA 7 via the interface 6.
[0045] The FPGA 7 may include a rewritable circuit area 70, and executes a predetermined process by a logic circuit set in the circuit area 70. Hereinafter, when distinguishing between the FPGAs 7, they will be represented as FPGAs #0 to #3 (see FIG. 3). For example, the FPGAs #0 to #3 may execute the above-mentioned processes A to D, respectively. In FIG. 3, the circuit scale of each process is indicated by reference numeral 71.
[0046] The optimization device 4 is an example of an information processing device or a computer, and is an optimizer that improves the processing efficiency of the FPGA 7. The optimization device 4 may be provided, for example, between the host PC 3 and the FPGA 7 so as to be able to communicate with each of them.
[0047] For example, the optimization device 4 may include a logic pool 40. The logic pool 40 is an example of a storage area, and may store a plurality of circuit logic templates 5 for which synthesis of a circuit design has been completed and placement and wiring has been determined. The template 5 is information on the circuit logic, and may be, for example, a bit stream.
[0048] When the optimization device 4 detects a load imbalance between a plurality of processes executed by a circuit set in a rewritable circuit area 70 included in each of one or more FPGAs 7, the optimization device 4 reads a first template that satisfies a preset condition from the logic pool 40. The first template may be, for example, a template 5 of a circuit that executes a process with a high load (as an example, a stalled process).
[0049] The optimization device 4 writes the read first template 5 to a first circuit area 70 that satisfies a condition among one or more circuit areas 70, thereby setting a first process realized by the circuit logic of the first template 5 in the first circuit area 70.
[0050] In recent years, FPGA7 has integrated various functions such as IO, network, etc. in addition to the dedicated functions. If the logic circuit to be written is prepared, only the dedicated function part of the functions mounted on FPGA7 can be rewritten in a few milliseconds, which has a wide range of applications.
[0051] In one embodiment, focusing on this point, by preparing in advance a template 5 of a circuit that can be set in the FPGA 7, the optimization device 4 can set a logic circuit in the circuit area 70 in real time or nearly in real time in order to eliminate the imbalance in the load between processes. This makes it possible to optimize the processes to be executed by the FPGA 7, and improve the processing efficiency of the FPGA 7.
[0052] [C] Hardware configuration example 4 is a block diagram showing an example of a hardware configuration of the system 1 according to an embodiment of the present invention, with attention focused on the connection configuration between the devices of the system 1 shown in FIG.
[0053] As shown in FIG. 4, the camera 2 and the host PC 3, and the host PC 3 and the optimization device 4 may be connected to each other so as to be able to communicate with each other via a network such as a LAN (Local Area Network) or the Internet.
[0054] The interface 6 may be, for example, an interconnect conforming to standards such as PCIe (Peripheral Component Interconnect-express) or CXL (Compute Express Link). For example, the interface 6 may be an adapter (connector) compatible with PCIe or CXL. In addition to the adapter, the interface 6 may also include a PCIe switch or a CXL switch that switches communication (connection) between the host PC 3 and each of the one or more FPGAs 7.
[0055] Furthermore, as shown in FIG. 4, the host PC 3 and the FPGA 7, and the optimization device 4 and the FPGA 7 may be connected to each other so as to be able to communicate with each other via a high-speed communication path 1a that complies with the standard of the IO port of the FPGA 7.
[0056] At least one of the host PC 3 and the optimization device 4 according to an embodiment may be a physical server or a virtual server (VM: Virtual Machine). At least one of the functions of the host PC 3 and the optimization device 4 may be realized by one computer or two or more computers.
[0057] The functions of the host PC 3 and the optimization device 4 according to an embodiment may be realized by a computer having a similar hardware (HW) configuration. The HW configurations of the host PC 3 and the optimization device 4 will be described below by taking the optimization device 4 as a representative example.
[0058] Figure 4 shows an example of a case where one computer is used as a HW resource to realize the functions of the optimization device 4, but if multiple computers are used, each computer may be equipped with the HW configuration shown in Figure 4.
[0059] As shown in FIG. 4, the optimization device 4 may include, as a HW configuration, a processor 4a, a graphics processing unit 4b, a memory 4c, a storage unit 4d, an IF (Interface) unit 4e, an IO unit 4f, and a reading unit 4g, for example.
[0060] The processor 4a is an example of a calculation processing device that performs various controls and calculations. The processor 4a may be connected to each block in the optimization device 4 via a bus 4j so that they can communicate with each other. The processor 4a may be a multiprocessor including a plurality of processors, a multicore processor having a plurality of processor cores, or a configuration having a plurality of multicore processors.
[0061] The processor 4a may be, for example, an integrated circuit (IC) such as a CPU, an MPU, an APU, a DSP, an ASIC, or an FPGA. The processor 4a may be a combination of two or more of these integrated circuits. MPU is an abbreviation for Micro Processing Unit. APU is an abbreviation for Accelerated Processing Unit. DSP is an abbreviation for Digital Signal Processor.
[0062] The graphics processing device 4b controls screen display for an output device such as a monitor in the IO unit 4f. The graphics processing device 4b may also have a configuration as an accelerator that executes machine learning processing and inference processing using a machine learning model. Examples of the graphics processing device 4b include various arithmetic processing devices, such as integrated circuits (ICs) such as GPUs (Graphics Processing Units), APUs, DSPs, ASICs, and FPGAs.
[0063] The memory 4c is an example of hardware that stores various types of data, programs, and other information. Examples of the memory 4c include a volatile memory such as a dynamic random access memory (DRAM) and / or a non-volatile memory such as a persistent memory (PM).
[0064] The storage unit 4d is an example of hardware that stores various data, programs, and other information. Examples of the storage unit 4d include various storage devices such as a magnetic disk device such as a hard disk drive (HDD), a semiconductor drive device such as an SSD, and a non-volatile memory. Examples of the non-volatile memory include a flash memory, a storage class memory (SCM), and a read only memory (ROM).
[0065] The storage unit 4d may store a program 4h (control program) that realizes all or part of the various functions of the optimization device 4.
[0066] For example, the processor 4a of the optimization device 4 can realize the function of the optimization device 4 (e.g., the control unit 48 shown in FIG. 5) described below by expanding the program 4h stored in the storage unit 4d into the memory 4c and executing it.
[0067] The IF unit 4e is an example of a communication IF that controls the connection and communication between the host PC 3 and the optimization device 4, between the optimization device 4 and the FPGA 7, etc. For example, the IF unit 4e may include an adapter that complies with various communication standards such as Ethernet (registered trademark), Infiniband, Millinet, PCIe, CXL, and an IO port of the FPGA 7. The adapter may support one or both of wireless and wired communication methods. The adapter may also be compliant with optical communication such as FC (Fibre Channel).
[0068] The program 4h may be downloaded to the optimization device 4 from a network (not shown) via the communication IF and stored in the storage unit 4d.
[0069] The IO unit 4f may include one or both of an input device and an output device. Examples of the input device include a keyboard, a mouse, a touch panel, etc. Examples of the output device include a display device such as a monitor, a projector, a printer, etc. The IO unit 4f may also include a touch panel or the like in which the input device and the output device are integrated. The output device may be connected to the graphics processing device 4b.
[0070] The reading unit 4g is an example of a reader that reads out data or program information recorded on the recording medium 4i. The reading unit 4g may include a connection terminal or device to which the recording medium 4i can be connected or inserted. Examples of the reading unit 4g include an adapter that complies with USB (Universal Serial Bus) or the like, a drive device that accesses a recording disk, and a card reader that accesses a flash memory such as an SD card. The recording medium 4i may store a program 4h, and the reading unit 4g may read the program 4h from the recording medium 4i and store it in the storage unit 4d.
[0071] Examples of the recording medium 4i include non-transitory computer-readable recording media such as magnetic / optical disks and flash memories. Examples of the magnetic / optical disks include flexible disks, CDs (Compact Discs), DVDs (Digital Versatile Discs), Blu-ray Discs, and HVDs (Holographic Versatile Discs). Examples of the flash memory include semiconductor memories such as USB memories and SD cards.
[0072] The above-mentioned hardware configuration of the optimization device 4 is an example. Therefore, the hardware in the optimization device 4 may be appropriately increased or decreased (for example, adding or deleting any block), divided, integrated in any combination, or added or deleted buses. For example, the optimization device 4 may omit the IO unit 4f and the reading unit 4g.
[0073] Moreover, the host PC 3 may have the same HW configuration as the optimization device 4. For example, the processor 4a of the host PC 3 can realize the function of the host PC 3 by expanding the program 4h stored in the storage unit 4d into the memory 4c and executing it.
[0074] [D] Software configuration example Fig. 5 is a block diagram showing an example of a software configuration of the optimization device 4 according to an embodiment. As shown in Fig. 5, the optimization device 4 may exemplarily include a memory unit 41, a communication unit 42, an information collection unit 43, a determination unit 44, a read unit 45, a write unit 46, and an update unit 47. The communication unit 42, the information collection unit 43, the determination unit 44, the read unit 45, the write unit 46, and the update unit 47 are an example of a control unit 48.
[0075] The memory unit 41 is an example of a storage area, and stores various data used by the optimization device 4. The memory unit 41 may be realized, for example, by a storage area included in one or both of the memory 4c and the storage unit 4d (see FIG. 4) of the optimization device 4.
[0076] 5, the memory unit 41 may be capable of storing, for example, a logical pool area 41a, judgment condition management information 41b, FPGA management information 41c, and template management information 41d. The judgment condition management information 41b, FPGA management information 41c, and template management information 41d are examples of information for managing judgment conditions, FPGAs 7, and templates 5, respectively.
[0077] The logical pool area 41a is a storage area secured as the logical pool 40 (see FIG. 3). Of the logical pool area 41a, the storage area that stores a plurality of templates 5 as the logical pool 40 may be realized by the storage area of the storage unit 4d. Also, of the logical pool area 41a, the storage area that stores the template 5 selected from the logical pool 40 (read from the storage unit 4d) to be written to the FPGA 7 may be realized by the storage area of the memory 4c.
[0078] The following describes the templates 5 stored in the logical pool 40. For example, assume that the host PC 3 (AI processing engine 30) collects captured images from a camera 2 used for autonomous driving or the like and performs AI processing.
[0079] In order to absorb the performance difference between the multiple cameras 2 and the difference between the captured images due to the imaging environment such as day and night, pre-processing is performed on the captured images. Pre-processing varies depending on the content of the AI processing, such as thinning, brightness correction, projective transformation, distortion correction, edge extraction, etc.
[0080] These processes are executed according to the maximum resolution. Therefore, a processing pipeline is formed for each resolution, and the FPGA 7 executes the processing for each processing pipeline. The processing included in the processing pipeline, in other words, the processing executed by the FPGA 7, includes, for example, the following series of processes. Processing pipeline: thinning → brightness correction → distortion processing → edge extraction
[0081] The number of input images (channels) that can be processed by one processing pipeline decreases as the resolution increases. The resolution of the captured image input to pre-processing can vary, for example, in descending order, from 8K (7680 x 4320), 4K (3840 x 2160), WQHD (2560 x 1440), FHD (1920 x 1080), WXGA++ (1600 x 900), and HD (1280 x 720). The numbers in parentheses for each resolution indicate the number of pixels horizontally and vertically.
[0082] In this way, the higher the resolution, the more the number of processing pipelines increases, and therefore the higher the resolution, the larger the circuit scale of the logic circuit written in the circuit area 70 of the FPGA 7. In the following explanation, of the processing pipelines (processing A to processing C) executed by each of the FPGAs #0 to #2 shown in Fig. 4, processing A is for 4K, processing B is for WQHD, and processing C is for FHD.
[0083] The logic pool 40 may store in advance a bit stream (step P43 in FIG. 2) generated upon completion of synthesis (step P2 in FIG. 2) and placement and wiring (step P42 in FIG. 2) for each of a plurality of processes including these processes A to C. In FIG. 3, templates 5 of logic circuits for executing each of processes A to G are stored in the logic pool 40.
[0084] The communication unit 42 performs various communications with the host PC 3 and the FPGA 7 using the IF unit 4e (see FIG. 4).
[0085] The information collecting unit 43 collects information from the host PC 3 and the FPGA 7 via the communication unit 42 .
[0086] For example, the information collecting unit 43 may use a system monitoring tool to monitor the operating status of the FPGA 7 via the communication path 1a. As an example, the information collecting unit 43 may acquire the network transfer rate of each FPGA 7 by the system monitoring tool. The acquisition of the network transfer rate of each FPGA 7 by the system monitoring tool may be performed, for example, by measuring the amount of information (data) transmitted by the information collecting unit 43 (and the determination unit 44 described later) to each FPGA 7. Then, the information collecting unit 43 may calculate the processing rate of each FPGA 7 by comparing the acquired network transfer rate with the maximum network transfer rate when each FPGA 7 operates with the maximum number of input videos (number of channels).
[0087] The throughput calculated by the information collecting unit 43 is an example of a load, and an example of a monitoring result of an operation status. For example, the throughput may be a ratio of a current network transfer rate to a maximum network transfer rate. The maximum network transfer rate may be measured in advance. The network transfer rate of each FPGA 7 may be obtained at predetermined time intervals (e.g., every 5 seconds) by a system monitoring tool.
[0088] Up to this point, an example of calculating the processing amount of the FPGA 7 has been described on the assumption that a logic circuit that executes a single type of processing (any of processing A to processing C: see FIG. 3) is set in each FPGA 7.
[0089] According to a method according to an embodiment, a logic circuit that executes two or more types of processing may be set in each FPGA 7. In such a case, the information collecting unit 43 may acquire the network traffic for each type of processing (template 5) that runs on the FPGA 7. The network traffic for each type of processing may be, for example, the network traffic of the logic circuit that executes the processing.
[0090] For example, the information collecting unit 43 may calculate the ratio of the network transfer rate for each type of acquired processing to the maximum network transfer rate when each processing operates with the maximum number of input videos (number of channels) as the processing rate for each processing. In this way, the information collecting unit 43 can estimate an accurate load based on the actual operating status of the logic circuit by using the network transfer rate to calculate the processing rate.
[0091] Also, for example, the information collecting unit 43 may acquire (receive) an execution request for preprocessing from the host PC 3. As an example, the information collecting unit 43 may acquire the execution request transmitted from the host PC 3 via the communication path 1a by the IF unit 4e (communication IF). Note that, if the host PC 3 can change the transmission mode of the execution request, it may transmit the execution request to the optimization device 4 via a network such as a LAN. In this case, the information collecting unit 43 may acquire the execution request by the IF unit 4e (communication IF) via the network.
[0092] The information collecting unit 43 may use the number of execution requests received from the host PC 3 to calculate the processing amount of the FPGA 7. As an example, the information collecting unit 43 may calculate, as the processing amount of the FPGA 7, the ratio of the number of execution requests issued from the host PC 3 to the FPGA 7 during a predetermined period to the maximum number of execution requests that the FPGA 7 can process during the predetermined period.
[0093] The judgment unit 44 judges whether an imbalance has occurred between multiple processes in the FPGA group based on the monitoring results of the operating status of the FPGA 7, and if an imbalance between multiple processes is detected, judges that a template 5 that meets predetermined conditions should be set in a circuit area 70 that meets the predetermined conditions.
[0094] The predetermined condition is a condition according to a policy that matches the state indicated by the monitoring result of the operational status among a plurality of policies.
[0095] Fig. 6 is a diagram showing an example of the judgment condition management information 41b. As shown in Fig. 6, the judgment condition management information 41b may include items of "condition name", "judgment condition (policy)", and "details".
[0096] The "condition name" is an example of identification information for an entry of a judgment condition. The "judgment condition (policy)" is various approaches for resolving processing imbalances, such as methods (perspectives) such as "efficiency" and "power saving." The "details" are specific conditions for achieving the approach of the judgment condition, and may define, for example, at least one of the following: the state of imbalance (detection condition), information on the template 5 to be written (reading process condition), and information on the FPGA 7 to which the template 5 is written (write destination condition).
[0097] The determination unit 44 may compare the monitoring result with the determination condition management information 41b, and when the monitoring result matches the detection condition, determine to write the template 5 that matches the condition of the read process of the entry to the FPGA 7 that matches the write destination condition. Then, the determination unit 44 notifies the reading unit 45 and the writing unit 46 of the determination result.
[0098] Fig. 7 is a diagram for explaining an example of the determination process by the determination unit 44. As indicated by reference symbol A1 in Fig. 7, it is assumed that, among FPGAs #0 to #2, the processing amount of FPGA #0 that executes process A is 50% or less, and the processing amounts of FPGAs #1 and #2 are on the rise. Process B or process C executed by FPGA #1 or #2 is an example of a first process, and process A executed by FPGA #0 is an example of a second process.
[0099] In the above judgment conditions, 50% is an example of a first threshold value. The increasing tendency may mean, for example, that the processing amount calculated this time by the information collecting unit 43 has increased by more than the second threshold value from the processing amount (for example, the average or weighted average) calculated in the past m times (m is an integer of 1 or more). The second threshold value may be a ratio of several percent to several tens of percent, or may be a predetermined value.
[0100] In this case, the judgment unit 44 judges that the process B of FPGA#1 or the process C of FPGA#2 should be written to the second logical partition of the circuit area 70 of FPGA#0 according to the entry (condition name) R1 or R2 of the judgment condition management information 41b. The circuit area 70 (second logical partition) of FPGA#0 is an example of a first circuit area. By such a judgment, the template 5 of the process B or the process C, whose processing amount is on the increase (whose processing is stalled), can be set in the circuit area 70 of FPGA#0, which has free processing resources (see the change from symbol A2 to symbol A3 in FIG. 7).
[0101] If there are multiple entries in the judgment condition management information 41b whose monitoring results match the detection condition, the judgment unit 44 may select the policy with the highest priority according to the priority of the policy. Examples of the priority of the policy (descending order) include descending order of the entry (condition name), descending order of the strictness of the detection condition (large number of conditions, low ratio of processing amount, etc.), or a designation by the administrator in advance.
[0102] The determination unit 44 may be realized, for example, by a function of an application such as an optimization solver, or may be realized by a machine learning model trained by a method such as deep learning. The machine learning model may be trained, for example, to output read processing conditions and write destination conditions according to detection conditions when a monitoring result is input.
[0103] Note that, when the system 1 (or the FPGA 7) is started up, the determination unit 44 may identify the template 5 to be initially written to each FPGA 7, and notify the reading unit 45 and the writing unit 46. The template 5 to be initially written may be designated by an administrator or a user of the system 1, or a determination condition for selecting the template 5 to be initially written may be set in the determination condition management information 41b.
[0104] Furthermore, the determination unit 44 refers to the FPGA management information 41c and the template management information 41d, and assigns the FPGA 7 to which an execution request sent from the host PC 3 is to be transferred to an appropriate FPGA 7 that processes the execution request.
[0105] Fig. 8 is a diagram showing an example of the FPGA management information 41c. As shown in Fig. 8, the FPGA management information 41c may include items such as "FPGA number", "number of logical partitions", "logical partition size", and "operating template number".
[0106] The "FPGA number" is an example of identification information of the FPGA 7. The "number of logical partitions" is the number of logical partitions set in the circuit area 70 of the FPGA 7. The logical partitions may be obtained by logically dividing the circuit area 70 at a timing such as the start-up of the FPGA 7, or may be dynamically set according to the number and size of processes (templates 5 to be written) set in the circuit area 70.
[0107] The "logical partition size" is the area size of the logical partition part. The area size may be measured in any unit. For example, the logical partition size may be an index relating to the ratio of the area size of the logical partition part to the area size of the circuit area 70, an index relating to the actual area of the logical partition part (e.g., the number of LBs (Logical Blocks) included in the logical partition), or the like. In the example of FIG. 8, the maximum value of the logical partition size per FPGA 7 (i.e., the area size of the circuit area 70) is 200.
[0108] The “operating template number” is an example of identification information of the template 5 set in the logical partition of the circuit area 70, and indicates the template 5 corresponding to the process operating in the FPGA 7.
[0109] Fig. 9 is a diagram showing an example of template management information 41d. As shown in Fig. 9, the template management information 41d may include items of "template number", "logical partition size", "logical partition size", and "operating template number".
[0110] The "template number" is an example of identification information of the template 5. The "logical partition size" is the area size that a logical circuit realized by the template 5 occupies in the circuit area 70 (logical partition), and is an example of the circuit scale 71. The unit of the area size may be the same as the logical partition size of the FPGA management information 41c.
[0111] The "processing content" is the content of the processing that is realized as a logic circuit by the circuit logic of template 5. For example, template #0 shown in Fig. 9 is template 5 of a logic circuit (size: 100) that realizes 4K processing.
[0112] For example, the determination unit 44 may identify a template 5 that realizes a process requested by an execution request transmitted from the host PC 3 from the template management information 41d, and identify an FPGA 7 on which the identified template 5 operates from the FPGA management information 41c. Then, the determination unit 44 may issue (transmit) the execution request transmitted from the host PC 3 to the identified FPGA 7 via the communication path 1a. This makes it possible to distribute the load for each type of template 5 (logic circuit), and improve the processing efficiency of the system 1 as a distributed processing system.
[0113] When the same template 5 is set in multiple logical partitions within one FPGA 7 or among multiple FPGAs 7, the processing of the template 5 can be executed in each of the multiple logical partitions. When the determination unit 44 identifies multiple FPGAs 7 (logical partitions) as candidates for the destination of the acquired execution request, the determination unit 44 may determine the destination of the execution request in accordance with the processing load of the FPGA 7 (logical partition).
[0114] For example, the determination unit 44 may manage a part of the storage area of the memory 4c of the optimization device 4 as a queue (buffer) that stores execution requests. The queue may be used in common for all execution requests, or may be provided for each template 5 or each FPGA 7. For example, the determination unit 44 may read out the execution requests stored in the queue using logic such as FIFO (First-In First-Out), and issue the execution request to an FPGA 7 (logical partition) selected from among candidate FPGAs 7 (logical partitions) to which the execution request is to be issued. The execution result of the process by the FPGA 7 is transmitted to the host PC 3 via the interface 6.
[0115] Furthermore, the judgment unit 44 may update the information 41b to 41d stored in the memory unit 41 after the writing unit 46 described later writes the template 5 to the circuit area 70, and after the update unit 47 described later updates the template 5 in the logical pool 40.
[0116] For example, after writing template 5 to the circuit area 70, the determination unit 44 may update the “operating template number” of the entry for the FPGA 7 in the FPGA management information 41c. Note that, if the logical partition changes according to template 5, the determination unit 44 may update the “number of logical partitions” and the “logical partition size” of the entry.
[0117] Furthermore, for example, when a template 5 is added to the logical pool 40, the determination unit 44 may add an entry for the added template 5 to the template management information 41d. Furthermore, when a template 5 is changed or deleted from the logical pool 40, the determination unit 44 may update or delete the corresponding entry in the template management information 41d.
[0118] Furthermore, for example, when the "details" of the judgment condition management information 41b change due to addition, modification, or deletion of the template 5, the judgment unit 44 may update each entry of the judgment condition management information 41b so as to conform to the change. The judgment condition management information 41b may be updated based on a judgment condition update request, which will be described later, for example.
[0119] The reading unit 45 reads the template 5 identified by the determination unit 44 from the logical pool 40. For example, the reading unit 45 reads the template 5 identified by the determination unit 44 from the logical pool 40 implemented by a partial storage area of the storage unit 4d, and writes it to the memory 4c.
[0120] The writing unit 46 writes the template 5 written in the memory 4c by the reading unit 45 to the circuit area 70 of the FPGA 7 identified by the determination unit 44 via the communication path 1a. As a method for writing the template 5 to the circuit area 70 by the writing unit 46, for example, a method similar to that used for downloading a bit stream to the circuit area 70 (step P8 in FIG. 2) may be adopted. When the writing of the template 5 is completed, the writing unit 46 notifies the determination unit 44 of the completion.
[0121] The update unit 47 updates the template 5 stored in the logical pool 40 based on the template update request. The template update request is an example of a registration request for a template 5, and may include a storage request for adding a new template 5, a change request for changing an existing template 5, a deletion request for deleting an existing template 5, and the like.
[0122] For example, when the updating unit 47 receives a storage request including information of a new template 5, it may store the template 5 in the logical pool 40. Furthermore, for example, when the updating unit 47 receives a change request including information of the changed template 5 and information identifying the original template 5, it may update the original template 5 in the logical pool 40 to the changed template 5. Furthermore, for example, when the updating unit 47 receives a deletion request including information identifying the template 5 stored in the logical pool 40, it may delete the template 5 from the logical pool 40. When the processing in response to the template update request is completed, the updating unit 47 notifies the determination unit 44 of the completion.
[0123] For example, the update unit 47 may receive a template update request from the host PC 3 or a computer (not shown) via a network such as a LAN. Alternatively, the update unit 47 may acquire a template update request input by an administrator or user of the system 1 via the IO unit 4f or the reading unit 4g.
[0124] The template update request may also include a judgment condition update request for updating judgment conditions that change with the update of the template 5. The judgment condition update request is an example of a condition registration request. The judgment condition update request may include information regarding modification or deletion of existing conditions included in the judgment condition management information 41b, or information regarding conditions to be newly added to the judgment condition management information 41b. The judgment condition update request may be transmitted from the host PC 3 or a computer not shown, independently of the template update request, or may be input via the IO unit 4f or the reading unit 4g.
[0125] In this way, the update unit 47 can appropriately update the template 5 or the judgment conditions, so that the process executed by the FPGA 7 can be flexibly changed (adapted) to a change in the environment of the system 1, such as a change in the camera 2. This can improve the operating efficiency of the FPGA 7, making it possible to provide low-cost or high-performance services to users of the system 1.
[0126] [E] Example of operation Next, an operation example of the system 1 according to an embodiment will be described. Fig. 10 is a sequence diagram illustrating an operation example of the system 1 according to an embodiment.
[0127] 10, the information collecting unit 43 and the judging unit 44 of the optimization device 4 judge whether the processing amount of the FPGA 7 based on the network transfer amount of each FPGA 7 matches the detection condition of the judgment condition management information 41b (process S1b). The information collecting unit 43 may calculate the processing amount based on the number of execution requests received from the host PC 3 (for example, the number of execution requests for each FPGA 7) (process S1a). Process S1a and / or process S1b may be executed periodically, for example, every 5 seconds.
[0128] When the information collection unit 43 and the determination unit 44 receive a pre-processing execution request from the host PC 3 (processes S2, S5, S8), they identify templates #0 to #2 corresponding to processes A to C indicated by each execution request based on the template management information 41d. The information collection unit 43 and the determination unit 44 identify FPGAs #0 to #2 in which the identified templates #0 to #2 are set based on the FPGA management information 41c, and transmit the respective execution requests to the identified FPGAs #0 to #2 via the communication path 1a (processes S3, S6, S9).
[0129] The FPGAs #0 to #2 transmit (response) the execution results of processes A to C in response to the execution requests to the host PC 3 via the interface 6 or the communication path 1a (processes S4, S7, and S10). This completes the process when the execution request for the pre-processing is issued from the host PC 3.
[0130] Next, as shown in process S11a and / or S11b, it is assumed that the processing amount of the FPGA 7 matches condition R1 of the judgment condition management information 41b in the judgment by the information collecting unit 43 and the judging unit 44. In this case, the judging unit 44 notifies the reading unit 45 and the writing unit 46 of condition R1.
[0131] The reading unit 45 reads the template #1 from the logical pool 40 (eg, the storage unit 4d) in accordance with the condition R1 (processing S12), and stores the read template #1 in the memory 4c (processing S13).
[0132] The writing unit 46 writes the template #1 stored in the memory 4c to the second logical partition in the circuit area 70 of FPGA #0 via the communication path 1a in accordance with the condition R1 (process S14), and notifies the determination unit 44 of the completion of the writing. Note that in the first logical partition in the circuit area 70 of FPGA #0, the template #0 (4K processing) continues to operate without being affected by the writing of the template #1 in the second logical partition.
[0133] The determination unit 44 updates the FPGA management information 41c based on the result of the writing unit 46 writing the template #1 into the FPGA #0 (processing S15).
[0134] Fig. 11 is a diagram showing an example of updating the FPGA management information 41c. As shown in Fig. 11, the determination unit 44 updates the logical partition size of the second logical partition of FPGA #0 from 100 to 70, and updates the active template number from template #0 to template #1.
[0135] With the above, the process of writing the template 5 to the FPGA 7 (the process when the determination condition is satisfied) is completed.
[0136] Next, a case will be considered in which a template update request is input by the user to the optimization device 4 via the IO unit 4f or the reading unit 4g.
[0137] When the update unit 47 receives a template update request (processing S16), it adds, changes, or deletes the template 5 from the logical pool 40 in accordance with the received template update request (processing S17), and notifies the judgment unit 44 that the update of the template 5 has been completed.
[0138] The judgment unit 44 updates the judgment condition management information 41b based on the result of updating the template 5 for the logical pool 40 by the update unit 47 (process S18).
[0139] 12 is a diagram showing an example of updating the template management information 41d. For example, assume that a new template #x is stored in the logical pool 40 in response to a template update request. Template #x is template 5 of a logical circuit that performs pre-processing (HD processing) on a captured image with HD resolution, and the logical partition size is 25. Based on this information, the determination unit 44 registers an entry for template #x in the template management information 41d.
[0140] This completes the update process for the template 5. If a request to update the judgment conditions is input to the optimization device 4 together with the template update request, the judgment unit 44 further updates the template management information 41d in process S18.
[0141] 13 is a diagram showing an example of updating the judgment condition management information 41b. For example, assume that a judgment condition update request including a condition Ry for writing the processing of template #x to the third logical partition of FPGA #0 when the processing amount of FPGA #0 is 75% or less is input to the optimization device 4 together with a template update request. Based on this information, the judgment unit 44 registers an entry for the condition name Ry in the judgment condition management information 41b.
[0142] [F] Other The technology according to the above-described embodiment can be modified and changed as follows.
[0143] For example, the functional blocks 42 to 47 included in the optimization device 4 shown in Fig. 5 may be merged in any combination or may be divided. The information 41b to 41d stored in the memory unit 41 shown in Fig. 5 may be merged in any combination or may be divided.
[0144] In addition, in one embodiment, the host PC 3 and the optimization device 4 are described as being separate computers, but this is not limited thereto. For example, the host PC 3 and the optimization device 4 may be a single computer, or at least some of the functions or information provided in the optimization device 4 may be provided in the host PC 3.
[0145] Furthermore, although the logical pool 40 is provided in the optimization device 4, the present invention is not limited to this and the logical pool 40 may be provided in a computer or storage device different from the optimization device 4. In this case, the reading unit 45 may read the template 5 from the logical pool 40 via the IF unit 4e, and the update unit 47 may update the template 5 of the logical pool 40 via the IF unit 4e.
[0146] In addition, in the embodiment, the FPGA 7 is taken as an example of a programmable logic circuit, but the present invention is not limited thereto, and the method according to the embodiment can be applied to a case where a CPLD (Complex PLD) is used. Note that the template 5 of the CPLD may be bit stream data in which the placement and wiring (which may be called fitting) of the circuit logic has been completed, similar to the template 5 of the FPGA 7.
[0147] Furthermore, in the embodiment, an example (see Figs. 7 and 11) has been described in which templates 5 for processing of two different resolutions are set in one FPGA 7, but the present invention is not limited to this. For example, by preparing a more detailed template 5 for each process and arranging it in the circuit area 70, it is possible to reduce unused areas in the circuit area 70 and use the circuit area 70 efficiently.
[0148] In addition, in the embodiment, the judgment condition included in the judgment condition management information 41b is a specific judgment condition that specifies the FPGA 7, but is not limited to this. For example, the judgment condition may be, for example, "write the process executed by the FPGA 7 with the largest processing amount to the FPGA 7 with the processing amount equal to or less than a predetermined value."
[0149] Furthermore, in the embodiment, an example is given in which the processing amount per FPGA 7 is used as the judgment condition, but the present invention is not limited to this. For example, the processing amount for each type of template 5 operating in the FPGA 7 may be used as the judgment condition. As an example, in a case where multiple types of templates 5 are operating in the FPGA 7, if the number of the first type of templates 5 is N and the processing amount (%) of the process executed by the first type of template 5 is Z, the judgment condition management information 41b may include the following judgment conditions.
[0150] If Z≦100*(N-1) / N (detection condition), write template 5 of a type with an increasing processing volume (read processing condition) to Z*N / 100 logical partitions of the first type of template 5 (write destination condition).
[0151] By using such a determination condition, it is possible to eliminate the imbalance in the processing load for each type of template 5, and to improve the processing efficiency of the FPGA 7.
[0152] In the above example, the processing volume of the same type of template 5 within one FPGA 7 is used as the judgment criterion, but this is not limited to this, and the processing volume for each type of template 5 within the entire group of FPGAs provided in the system 1 may also be used.
[0153] Also, in the embodiment, an example has been described in which a maximum of two logic circuits for executing process A (4K process) can be written in the circuit area 70 of the FPGA 7. If the area size of the circuit area 70 increases with further integration density of the FPGA 7, a large number of logic circuits will be set in the circuit area 70. In this case, if the template 5 is repeatedly written in the circuit area 70 by the method according to the embodiment, logic circuits of various sizes may be discretely arranged in the circuit area 70 in units of template 5. As a result, even if there is a sufficient unused area (free area) in the circuit area 70, there is a risk that there will not be enough area for setting logic circuits.
[0154] Therefore, the judgment unit 44 may, for example, eliminate the discrete placement of template 5 units by repeating the process of rearranging the templates 5 set in the FPGA 7 or other FPGAs 7 in the logical partition of the FPGA 7 with a reduced processing volume.
[0155] Alternatively, when multiple systems 1 (or FPGA groups) exist, unused circuit areas 70 of one system 1 (or FPGA group) may be utilized to eliminate the discrete placement of template 5 units in the other system 1 (or FPGA group).
[0156] [G] Notes The following supplementary notes are further disclosed regarding the above embodiment.
[0157] (Appendix 1) when detecting a load imbalance between a plurality of processes executed by a circuit set in a reprogrammable circuit area included in each of the one or more programmable logic circuits, reading out a first template that satisfies a preset condition from a storage area that stores a plurality of circuit logic templates for which synthesis of a circuit design has been completed and placement and wiring have been determined; writing the read first template into a first circuit area that satisfies the condition among the one or more circuit areas, thereby setting a first process that is realized by a circuit logic of the first template in the first circuit area; A control device having a control unit.
[0158] (Appendix 2) the condition is a condition indicating that a first process, the load of which is increasing, is to be written to the first circuit area that executes a second process, the load of which is equal to or less than a first threshold. 2. The control device according to claim 1.
[0159] (Appendix 3) the control unit calculates a load for each of the plurality of processes based on a network transfer rate of each of the plurality of logic circuits that execute the plurality of processes; 3. The control device according to claim 1 or 2.
[0160] (Appendix 4) the control unit, in response to receiving a registration request for a new template, registers the new template in the storage area. The control device according to any one of Supplementary Note 1 to Supplementary Note 3.
[0161] (Appendix 5) The control unit is acquiring a processing request issued from an information processing device to the one or more programmable logic circuits; transmitting the processing request to a programmable logic circuit in which a template that realizes the processing requested by the processing request operates, based on information that manages the processing realized by the circuit logic of the plurality of templates and the templates corresponding to the processing that operates on each of the one or more programmable logic circuits; The control device according to any one of Supplementary Note 1 to Supplementary Note 4.
[0162] (Appendix 6) The condition defines at least one of the detection condition for the imbalance, a template to be written, and a circuit area to which the template to be written is to be written; The control unit is Identifying a condition in which the load of each of the plurality of processes matches the detection condition from information managing the plurality of conditions; identifying the first template and the first circuit area that satisfy the identified condition; The control device according to any one of Supplementary Note 1 to Supplementary Note 5.
[0163] (Appendix 7) when detecting a load imbalance between a plurality of processes executed by a circuit set in a reprogrammable circuit area included in each of the one or more programmable logic circuits, reading out a first template that satisfies a preset condition from a storage area that stores a plurality of circuit logic templates for which synthesis of a circuit design has been completed and placement and wiring have been determined; writing the read first template into a first circuit area that satisfies the condition among the one or more circuit areas, thereby setting a first process that is realized by a circuit logic of the first template in the first circuit area; A control method in which processing is executed by a computer.
[0164] (Appendix 8) the condition is a condition indicating that a first process, the load of which is increasing, is to be written to the first circuit area that executes a second process, the load of which is equal to or less than a first threshold. 8. The control method according to claim 7.
[0165] (Appendix 9) The computer, Calculating a load for each of the plurality of processes based on a network transfer rate of each of a plurality of logic circuits that execute the plurality of processes; 9. The control method according to claim 7 or 8, further comprising:
[0166] (Appendix 10) The computer, In response to receiving a registration request for a new template, registering the new template in the storage area. The control method according to any one of Supplementary Note 7 to Supplementary Note 9, further comprising executing a process.
[0167] (Appendix 11) The computer, acquiring a processing request issued from an information processing device to the one or more programmable logic circuits; transmitting the processing request to a programmable logic circuit in which a template that realizes the processing requested by the processing request operates, based on information that manages the processing realized by the circuit logic of the plurality of templates and the templates corresponding to the processing that operates on each of the one or more programmable logic circuits; The control method according to any one of claims 7 to 10, further comprising executing a process.
[0168] (Appendix 12) The condition defines at least one of the detection condition for the imbalance, a template to be written, and a circuit area to which the template to be written is to be written; The computer, Identifying a condition in which the load of each of the plurality of processes matches the detection condition from information managing the plurality of conditions; identifying the first template and the first circuit area that satisfy the identified condition; The control method according to any one of claims 7 to 11, further comprising executing a process.
[0169] (Appendix 13) one or more programmable logic circuits; A control device, The control device includes: when detecting a load imbalance between a plurality of processes executed by a circuit set in a rewritable circuit area included in each of the one or more programmable logic circuits, reading out a first template that satisfies a preset condition from a storage area that stores a plurality of circuit logic templates for which synthesis of a circuit design has been completed and placement and wiring has been determined; a control unit that writes the read first template into a first circuit area that satisfies the condition among one or more of the circuit areas, thereby setting a first process that is realized by a circuit logic of the first template in the first circuit area; Distributed processing system.
[0170] (Appendix 14) the condition is a condition indicating that a first process, the load of which is increasing, is to be written to the first circuit area that executes a second process, the load of which is equal to or less than a first threshold. 14. The distributed processing system of claim 13.
[0171] (Appendix 15) the control unit calculates a load for each of the plurality of processes based on a network transfer rate of each of the plurality of logic circuits that execute the plurality of processes; 15. The distributed processing system according to claim 13 or 14.
[0172] (Appendix 16) the control unit, in response to receiving a registration request for a new template, registers the new template in the storage area. The distributed processing system according to any one of claims 13 to 15.
[0173] (Appendix 17) The control unit is acquiring a processing request issued from an information processing device to the one or more programmable logic circuits; transmitting the processing request to a programmable logic circuit in which a template that realizes the processing requested by the processing request operates, based on information that manages the processing realized by the circuit logic of the plurality of templates and the templates corresponding to the processing that operates on each of the one or more programmable logic circuits; The distributed processing system according to any one of claims 13 to 16.
[0174] (Appendix 18) The condition defines at least one of the detection condition for the imbalance, a template to be written, and a circuit area to which the template to be written is to be written; The control unit is Identifying a condition in which the load of each of the plurality of processes matches the detection condition from information managing the plurality of conditions; identifying the first template and the first circuit area that satisfy the identified condition; The distributed processing system according to any one of claims 13 to 17. [Explanation of symbols]
[0175] 1 System 1a Communication Channel 2 Camera 3. Host PC 30 AI processing engine 4 Optimization Device 40 logical pools 41 Memory section 41a Logical Pool Area 41b Judgment condition management information 41c FPGA management information 41d Template Management Information 42 Communications Department 43 Information Gathering Department 44 Judgment section 45 Reading section 46 Writing section 47 Update section 48 Control Unit 5. Templates 6 Interface 7 FPGA 70 circuit area 71 Circuit scale
Claims
1. When detecting a load imbalance between a plurality of processes executed by a circuit set in a rewritable circuit area included in each of the one or more programmable logic circuits, reading out a first template that satisfies a preset condition from a storage area that stores a plurality of circuit logic templates for which synthesis of a circuit design has been completed and placement and wiring has been determined; writing the read first template into a first circuit area that satisfies the condition among the one or more circuit areas, thereby setting a first process that is realized by a circuit logic of the first template in the first circuit area; A control device having a control unit.
2. the condition is a condition indicating that a first process, the load of which is increasing, is to be written to the first circuit area that executes a second process, the load of which is equal to or less than a first threshold. The control device according to claim 1 .
3. the control unit calculates a load for each of the plurality of processes based on a network transfer rate of each of the plurality of logic circuits that execute the plurality of processes; The control device according to claim 1 or 2.
4. the control unit, in response to receiving a registration request for a new template, registers the new template in the storage area. The control device according to claim 1 or 2.
5. The control unit is acquiring a processing request issued from an information processing device to the one or more programmable logic circuits; transmitting the processing request to a programmable logic circuit in which a template that realizes the processing requested by the processing request operates, based on information that manages the processing realized by the circuit logic of the plurality of templates and the templates corresponding to the processing that operates on each of the one or more programmable logic circuits; The control device according to claim 1 or 2.
6. The condition defines at least one of a detection condition for the imbalance, a template to be written, and a circuit area to which the template to be written is to be written; The control unit is Identifying a condition in which the load of each of the plurality of processes matches the detection condition from information managing the plurality of conditions; identifying the first template and the first circuit area that satisfy the identified condition; The control device according to claim 1 or 2.
7. When detecting a load imbalance between a plurality of processes executed by a circuit set in a rewritable circuit area included in each of the one or more programmable logic circuits, reading out a first template that satisfies a preset condition from a storage area that stores a plurality of circuit logic templates for which synthesis of a circuit design has been completed and placement and wiring has been determined; writing the read first template into a first circuit area that satisfies the condition among the one or more circuit areas, thereby setting a first process that is realized by a circuit logic of the first template in the first circuit area; A control method in which processing is executed by a computer.
8. one or more programmable logic circuits; A control device, The control device includes: when detecting a load imbalance between a plurality of processes executed by a circuit set in a rewritable circuit area included in each of the one or more programmable logic circuits, reading out a first template that satisfies a preset condition from a storage area that stores a plurality of circuit logic templates for which synthesis of a circuit design has been completed and placement and wiring has been determined; a control unit that writes the read first template into a first circuit area that satisfies the condition among one or more of the circuit areas, thereby setting a first process that is realized by a circuit logic of the first template in the first circuit area. Distributed processing system.
Citation Information
Patent Citations
Video image processing device and video image display device
JP2015184549A
Information processing apparatus having accelerator and information processing method
JP2018132981A
Systems and methods for aggregate bandwidth and latency optimization
US20190334771A1
Service deployment method and network functions acceleration platform
US20200099579A1
