Performance-aware fine-grained scheduling for accelerated resources in industrial containerized platforms

Abstract

Current approaches to deploying graphical processing units (GPUs) on edge devices lack efficiencies and capabilities. For example, current approaches often waste or mis-use resources, or fail to meet timeliness requirements of high priority applications, among other technical issues. An orchestration system can be configured to monitor and orchestrate resources in a fine-grained manner, so as to meet timeliness requirements of high priority applications while fully utilizing a given resource slack.

Description

PERFORMANCE- AW ARE FINE-GRAINED SCHEDULING FOR ACCELERATEDRESOURCES TN INDUSTRIAL CONTAINERIZED PLATFORMSBACKGROUND

[0001] Industrial automation systems and applications can be used to control the operation of machines and other components in a systematic manner. Automation systems can include various automation domains such as factory automation, process automation, building automation, energy automation, and the like. Various industrial applications, such as process monitoring, collision detection vibration monitoring, and production control can utilize machine learning-based approaches for various operational tasks. In many cases, due to low latency, privacy concerns, or data cost requirements, such tasks arc deployed on edge devices. Furthermore, general purpose graphics processing units (GPUs) are required to execute various machine learning applications in a timely fashion. Due to presence of several tasks in the current generation of industrial processes, these devices often need to host multiple tasks in parallel with different priorities. It is recognized herein, however, that current approaches to deploying tasks with different priorities executing on edge device’s GPU, lack efficiencies and capabilities. For example, current approaches often waste or under-utilize resources, or fail to meet timeliness requirements of high priority applications, among other technical issues.BRIEF SUMMARY

[0002] Embodiments of the invention address and overcome one or more of the described- herein shortcomings or technical problems by providing methods, systems, and apparatuses for monitoring and orchestrating resources in a fine-grained manner, so as to meet timeliness requirements of high priority applications while fully utilizing a given resource slack.

[0003] In an example aspect, an industrial network includes an orchestration system that defines a master node edge device and a plurality of worker node edge devices communicatively coupled to the master node edge device. The plurality of worker node edge devices defines a plurality of graphical processing units that each define a memory and a plurality of streaming multiprocessors. The orchestration system, in particular the master node edge device, can obtain a list of a plurality of virtual execution engines that each define anabstraction of the memory and a streaming multiprocessor of the plurality of streaming multiprocessors, such that each graphical processing unit of the plurality of graphical processing units defines a plurality of virtual execution engines. The system can further obtain a plurality of tasks to be performed by one or more of the plurality of graphical processing units. The system can determine an execution time associated with a first task of the plurality of tasks, and determine task deadlines associated with each of the virtual execution engines. Based on the execution time and the task deadlines, the master node edge device can assign the first task to a first virtual execution engine of the plurality of virtual execution engines for completion. For example, each task of the plurality of tasks can be associated with a priority, and the master node can select the first task from the plurality of tasks based on the priority of the first task.

[0004] Tn another example aspect, the system can retrieve a task profile associated with the first task, from a database, wherein the task profile indicates the execution time. To generate the task profile, the system can collect metrics while executing the first task on the plurality of virtual execution engines, so as to generate the task profile based on the metrics.

[0005] In some cases, before assigning the first task to the first virtual execution engine, the system can make a determination that the first virtual execution engine is available.Responsive to the determination, the system can assign the first task to the first virtual execution engine. Alternatively, the system can determine that none of the virtual execution engines in the list are available to process tasks. The system can further make a determination that the first virtual execution engine is scheduled to perform a scheduled task that has a lower priority than other scheduled tasks that the other virtual execution engines in the list are scheduled to perform. Responsive to such a determination, the system can assign the first task to the first virtual execution engine. Additionally, responsive to such a determination, the system, in particular the master node, can replace the scheduled task, from a schedule of the first virtual execution engine, with the first task. In various examples, the system can make a determination that the execution time associated with the first task of the plurality of tasks is less than the task deadline associated with the first virtual execution engine. Based on this determination, the master node edge device can assign the first task to the first virtual execution engine.BRIEF DESCRIPTION OF THE DRAWINGS

[0006] The foregoing and other aspects of the present invention are best understood from the following detailed description when read in connection with the accompanying drawings. For the purpose of illustrating the invention, there is shown in the drawings embodiments that are presently preferred, it being understood, however, that the invention is not limited to the specific instrumentalities disclosed. Included in the drawings are the following Figures:

[0007] FIG. 1 is a block diagram of an example industrial control network that includes a plurality of edge devices or nodes, in accordance with an example embodiment.

[0008] FIG. 2 is a block diagram of an example orchestration system that includes an edge device configured as a master node, and a plurality of edge devices that define graphical processing units (GPUs) configured as worker nodes, in accordance with an example embodiment.

[0009] FIG. 3 is block diagram that illustrates example modules of the worker node that communicate with each other to add awareness, in accordance with an example embodiment.

[0010] FIG. 4 is a flow diagram of operations that can be performed by the orchestration system so as to develop a model of the performance of the schedule tasks that are executed by the GPUs of the worker nodes.

[0011] FIG. 5 is a flow diagram of operations that can be performed by the orchestration system so as to schedule jobs that are performed by the GPUs of the worker nodes.

[0012] FIG. 6 shows an example of a computing environment within which embodiments of the disclosure may be implemented.DETAILED DESCRIPTION

[0013] As an initial matter, it is recognized herein that while graphical processing units (GPUs) are increasingly integrated into edge hardware platforms via PCIe ports, technical problems have not been addressed related to utilizing the capabilities of the hardware when multiples tasks need to share the GPU. For example, Linux Containers are often used to deploy applications on shared platforms due to low overhead and ease of deployment. It is recognizedherein, however, that state-of-the-art orchestration frameworks have limited management support for GPU acceleration for containers. Such limited support often results in dedicating GPUs to individual applications, which is often not practical as there can be multiple applications that require acceleration. Furthermore, such limited support can waste resources because a single application often cannot fully utilize a given GPU due to, for example, burstiness of workloads and limited memory bandwidth particularly on embedded accelerators. rooi4i It is recognized herein that the current generation of orchestrators (e.g., Kubernetes) lack fine-grained awareness of accelerator resources at the edge devices. Furthermore, such orchestrators often require vendor specific solutions for awareness of GPU resources, and such orchestrators cannot perform resource sharing or fractional allocation. It is further recognized herein that current orchestrators that support GPUs do not consider the priority, estimated execution time, or deadline of applications running on the hardware resources. Applications with different criticality levels, however, might need to execute concurrently on hardware platforms while guaranteeing performance to critical applications. Further still, it is recognized herein that current resource isolation techniques are coarse grained for embedded GPU devices, which can lead to resource wastage with limited support for partitioning a given GPU across multiple task kernels.

[0015] In contrast, in accordance with various embodiments described herein, a system includes a node agent that defines fine-grained resource awareness. Furthermore, the system can perform real-time GPU scheduling, so as to prioritize applications and increase resource sharing. As further described herein, container runtime is enhanced, so that scheduling can be performed at the cluster (e.g., group of GPUs or edge devices) and node (e.g., a single edge device) levels.

[0016] Referring initially to FIG. 1, an example automation or industrial control network or system 100 can include one or more plants or production networks 104 that contain control logic, host web servers, and the like. For example, the industrial control network 100 can include an enterprise or IT network 102 and multiple operational plant or production networks 104 communicatively coupled to the IT network 102. The production network 104 or enterprise network 102 can include a plurality of edge devices or nodes 106 connected within the production network 104. The edge devices 106 can include virtual programmable logic controllers (PLCs), industrial computers, management computing devices, or the like. The edge devices 106 that are configured as worker nodes (further described herein) can includeembedded or plugged GPUs, such as an industrial PC with GPU added. In various examples, an edge device 106 with or without a GPU can operate as the master node further described herein. Thus, various industrial PCs or commercial-off-the-shelf servers or single board computers can be configured as the master node. An example edge device 106 is connected to the IT network 102. The arrangement of edge devices 106 can vary as desired, and all such arrangements are contemplated as being within the scope of this disclosure. rooi7i Still referring to FIG. 1, the production network 104 can include various production machines configured to work together to perform one or more manufacturing operations. Example production machines of the production network 104 can include, without limitation, robots 108 and other field devices that can be controlled by a respective PLC 114, such as sensors 110, actuators 112, or other machines, such as automatic guided vehicles (AGVs). The PEC 114 can send instructions to respective field devices. In some cases, a given PEC 114 can be coupled to a human machine interfaces (HMIs) 1 16. It will be understood that the industrial control network 100 is simplified for purposes of example. That is, the industrial control network 100 may include additional or alternative nodes or systems, for instance other network devices, that define alternative configurations, and all such configurations are contemplated as being within the scope of this disclosure.

[0018] The network or system 100, in particular each production network 104, can define a field portion or level 118 and plant level or portion 120. For example, and without limitation, the plant level 120 can define one or more industrial plants or systems that can be geographically and functionally separate from or independent of each other. For example, the plant level 120 can include Brownfield plants and Greenfield plants that are each connected to respective field devices within the field level 118. The field level 118 can include various field devices such as the robots 108, PLC 114, sensors 110, actuators 112, HMIs 116, and AGVs. The field portion 118 can define one or more production lines or control zones associated with a given plant in the plant level 120. The PLC 114, sensors 110, actuators 112, and HMI 116 within a given production line can communicate with each other via a respective field bus 122. Each control zone can be defined by a respective PLC 114, such that the PLC 114, and thus the corresponding control zone, can connect to the respective plant portion 120 via an Ethernet connection 124. In some cases, the robots 108 and AGVs can be configured to communicate with other devices within the fieldbus portion 118 via a Wi-Fi connection 126. Similarly, the robots 108 and AGVs can communicate with the Ethernet portion 120, in particular aSupervisory Control and Data Acquisition (SCADA) server 128, via the Wi-Fi connection 126. In various examples, a respective edge device 106 is communicatively coupled between the PLC 114 and the respective plant in the plant level 120, for instance via the Ethernet connection 124 or the Wi-Fi connection 126.

[0019] The plant level 120 of a given production network 104 can include various computing devices or subsystems communicatively coupled together via the Ethernet connection 124. Example computing devices or subsystems in the plant portion 120 include, without limitation, a mobile data collector 130, HMIs 132, the SCADA server 128, the edge devices 106, a wireless router 134, a manufacturing execution system (MES) 136, an engineering system (ES) 138, and a log server 140. The ES 138 can include one or more engineering works stations. In an example, the MES 136, HMIs 132, ES 138, and log server 140 are connected to the production network 104 directly. The wireless router 134 can also connect to the production network 104 directly. Thus, in some cases, mobile users, for instance the mobile data collector 130 and robots 108 (c.g., AGVs), can connect to the production network 104 via the wireless router 134.

[0020] As described above, the industrial control network or automation system 100 can define a heterogenous IT / OT network, for instance an IT / OT network that includes the IT network 102 and the production network 104. For example, the IT network 102 and the production network 104 can include or implement various devices, protocols, users, network administrators, and the like so as to define a heterogenous IT / OT network. In various examples, the industrial control network 100, and thus the heterogeneous IT / OT network, defines multiple production networks 104 or plants across multiple sites, each with their own operational histories.

[0021] Referring now to FIG. 2, the example industrial system 100 can define an orchestration system 200 that includes a master node or master node edge device 202 communicatively coupled to a plurality of worker nodes or worker node edge devices 204. In various examples, the plurality of edge devices 106 can define the master node 202 and respective worker nodes 204. It will be understood that the number of worker nodes 204 can vary as desired, and all such numbers of worker nodes are contemplated as being within the scope of this disclosure.

[0022] The worker nodes 204 can each include one or more graphical processing units (GPUs) 206 and memory having stored thereon applications, agents, and computer program modules including, for example, a node agent 210, a container runtime module 212, and a plurality ofcontainers 214. Each GPU 206 can define a plurality of streaming multiprocessors (SM) 216 coupled to a cache 218 and a GPU memory 220. Each GPU 206 can further define a copy engine 222. The copy engine 222 can define a GPU module that can copy blocks of data from one memory location to another for processing. The streaming multiprocessors 216 and the GPU memory 220 can define a plurality of virtual execution engines (vEEs) 224. The master node 202 can include on or more processors and memory having stored thereon applications, agents, and computer program modules including, for example, an orchestrator or scheduler module 226, a controller manager 228, and an application programming interface (API) server 230 communicatively coupled to the scheduler module 226 and the controller manager 228. It will be appreciated that the program modules, applications, computer-executable instructions, code, or the like depicted in FIG. 2 are merely illustrative and not exhaustive, and that processing described as being supported by any particular module may alternatively be distributed across multiple modules or performed by a different module. In addition, various program module(s), script(s), plug-in(s), Application Programming Interface(s) (API(s)), or any other suitable computcr-cxccutablc code may be provided to support functionality provided by the program modules, applications, or computer-executable code depicted in FIG. 2 and / or additional or alternate functionality. Further, functionality may be modularized differently such that processing described as being supported collectively by the collection of program modules depicted in FIG. 2 may be performed by a fewer or greater number of modules, or functionality described as being supported by any particular module may be supported, at least in part, by another module. In addition, program modules that support the functionality described herein may form part of one or more applications executable across any number of systems or devices in accordance with any suitable computing model such as, for example, a client-server model, a peer-to-peer model, and so forth. In addition, any of the functionality described as being supported by any of the program modules depicted in FIG. 2 may be implemented, at least partially, in hardware and / or firmware across any number of devices.

[0023] Still referring to FIG. 2, the virtual execution engine vEE 224 can define an abstraction at the level of the stream multiprocessors 216. In various examples, each vEE 224 can consist of the structure in Table 1:Table 1

[0024] The vEE 224 can indicate a streaming multiprocessor ID (SMID), which can be a unique identity (ID) that can be mapped to a respective streaming multiprocessor (SM) 216 for execution. Each vEE 224 can also define a memory size, which is the size of the configurable memory of the respective vEE 224. The memory size can depend on the hardware and typical expected workload of the related SM 216. Each vEE 224 can further define a kernel ID that uniquely identifies the kernel to be executed, and a priority that indicates the priority of the kernel to be executed. In various examples, the lower the number of the priority, the higher the priority. Thus, the priority can be a characteristic of the kernel / task. In various examples, the priority is user-defined. In some cases, memory size is defined by equally dividing the GPU memory 220 across the streaming multiprocessors 216, though it will be understood that memory can be partitioned differently as desired.

[0025] The orchestration system 200 is aware of the virtual execution engines (vEEs) 224 defined by each worker node 204. In particular, referring also to FIG. 3, the node agent 210 (e.g., KubeCtl in Kebernetes) can define a mapper component 302 configured to translate the vEE 224 to one or more physical streaming multiprocessors 216 and memory 220. Similarly, the mapper component 302 can translate the streaming multiprocessor 216 and memory 220 to one or more vEEs 224. In an example, the mapper component 302 receives instructions from the orchestrator or scheduler module 226 running on the master node 202. Alternatively, or additionally, the mapper component 302 can have awareness of the underlying resources that it uses for assignments. By way of example, if the scheduler module 226 asks the mapper component 302 to assign two (2) streaming multiprocessors 226 and 1 GB of memory 220, the mapper component can communicate with the operating system running on the host (edge device or worker node 204) to determine if a match exists. The node agent 210 can further include a monitor component 304 configured to query the container runtime module 212, for a current state of the respective GPU 206. In some examples, the monitor component 304 can receive the state from an extended reporter module 306 periodically or on-demand when the orchestrator 226 needs to make a scheduling decision. For example, the container runtime module 212 can define the extended reporter module 306 (e.g., cAdvisor) that can be configured to, in response to the query, report the utilization of the respective GPU memory 220 and streaming multiprocessors 216. Additionally, the node agent 210 can further define anexecutor module 306 configured to specify the SM 216 and GPU memory 220 that is assigned to a given container that hosts the kernel. In particular, the container runtime module 212 can define a SM and memory aware container 310 that can receive the communication from the executor module 308 that indicates the assigned SM 216 and GPU memory 220. The executor module 306 can also preempt lower priority kernels based on scheduling decisions that are received from the master node 202, as further described herein.[00261 Referring now to FIGs. 2-5, the orchestration system 200 can perform example scheduling operations 400 and 500. In particular, the scheduler module 226, based on its awareness of the streaming multiprocessors 216 and information from Table 1 that it receives via the node agent 210, can schedule the GPUs 206 for concurrent execution of various operations (tasks) or multiple jobs. The controller manager 228 can manage the execution of those tasks, so as to maintain the state of the system 200. For example, based on the tasks that the GPUs 206 are performing, the scheduler module 226 can schedule available GPUs 206 for higher priority tasks, for instance running a virtual PLC, before scheduling GPUs 206 for lower priority tasks, for instance performing machine learning training. In some examples, the scheduler module 226 can preempt tasks that a given GPU 206 or cluster (group) of GPUs 206 is performing, so that high priority tasks or jobs can be executed instead.

[0027] Referring to FIG. 4 and FIG. 5, scheduling operations 400 can define a first phase of scheduling, and scheduling operations 500 can define a second or online phase of scheduling. In the first phase, with particular reference to FIG. 4, an application is profiled to establish the estimated execution time of the application. The operations 400 can define a one-time task that is performed for the set of edge device types available in a cluster of worker nodes. For example, at 402, a new application can be executed on the edge devices per vEE 224. Furthermore, at 402, the performance metrics and signature of the vEE utilization can be collected by the infrastructure of the orchestration system 200. At 404, the signature can be compared to a database of application profiles to retrieve any existing task profiles. At 406, the system 200 can determine whether the signature is unique, thereby determining whether the relevant task profile already exists. If the signature is unique, the system 200 determines that no relevant task profile exists, and the process can proceed to 408. At 408, the system 200 can deploy a set of pre-defined stressors that measure the performance of the application. For example, the stressors can vary the GPU 206 and memory 220 utilization from 0 to 100%.Based on the stressors, at 410, the system 200 can determine the estimated execution of theapplication to be used during runtime. For example, at 410, the system 200 can develop a task profile using the collected metrics for each vEE 224. At 412, the system 200 can add the task profile to the database, so as to update the task profile database. At 414, the task profile can be selected and used during the online phase of scheduling that includes operations 500. When, at 406, the system 200 determines that the signature is not unique, and thus the a relevant task profile exists, the process can proceed to 414, where the relevant existing task profile is selected.

[0028] Referring in particular to FIG. 5, at 502, the system 200 can retrieve a list of all the VEEs 224 and their categories, so as to compile the list of available resources. Such resources can be characterized by the categories, which can be defined by hardware and capacities. For example, hardware from different processing generations and with difference capacities can define different categories. At 504, the scheduler module 226 can actively listen, for instance to the APT server 230, to receive a new set of unscheduled tasks. Such unscheduled tasks can indicate various information associated with the tasks, such as user or system defined priorities, task deadlines, and associated requirements for memory 220 and streaming multiprocessors 216. At 506, the scheduler module 226 can sort the tasks in descending order of the user or system defined priorities, so as to define a sorted list of tasks. At 508, the task profile obtained from 414 can be used to estimate the application performance, for instance the execution time. At 510, the scheduler module 226 can retrieves the list of available vEEs 224 from the node agent 210. At 512, the scheduler module 226 can select the highest priority task from the sorted list. In some cases, at 514, the scheduler module 226 can determine that the task list is empty. In such cases, the scheduler module 226 can wait for new tasks to arrive, at 504. Alternatively, when the scheduler module 226 determines that the task list is not empty, the process can proceed to 516, where the scheduler module 226 can select an unallocated vEE 224. At 518, the scheduler module 226 can verify or determine that the selected vEE 224 is available. When it is determined that the selected vEE 224 is available, the scheduler module 226 can attempt to fit the task to the selected vEE 224, at 520. In particular, for example, the system 200 can determine whether the estimated execution time for the task is less than, or is completed before, the task deadline for the selected vEE 224, such that the selected vEE 224 can support the task. When it is determined that the selected vEE cannot support the task, the process can return to 516, where another vEE 224 is selected. When it is determined that theselected vEE can support the task, scheduler module 226 can assign the task to the selected vEE for completion, at 522.

[0029] Still referring to FIG. 5, when it is determined (at 518) that the selected vEE 224 is not available, the schedule module 226 can retrieve the list of scheduled vEEs 224 from the node agent 210, at 524. Furthermore, at 524, if no vEE 224 is available such that the list only includes scheduled vEEs 224, the scheduler module 226 can sort the scheduled vEEs 224 based on the priority of tasks that the respective vEE has scheduled and then based on amount of time executed. Thus, at 526, the scheduler module 226 can select a scheduled vEE 224, so as to preempt the scheduled vEE 224 associated with the lowest priority task. Alternatively, if the priorities are the same for multiple vEEs 224, the scheduled vEE 224 with the least executed task can be selected, at 526. At 528, the system determines if the target application can fit in the selected scheduled vEE 224. For example, the system can determine whether the estimated execution time for the application is less than the task deadline for the selected scheduled vEE. If the estimated execution time is greater than the task deadline, the process can return to 526, where the next scheduled vEE 224 in the sorted list is selected. If the estimated execution time is less than the task deadline, the task (e.g., lowest priority application) associated with the selected scheduled vEE 224 is removed (unscheduled) and put in the unscheduled task list, at 530. The target task can then be scheduled and assigned, at 522.

[0030] Thus, as described herein, an industrial network includes an orchestration system that defines a master node edge device and a plurality of worker node edge devices communicatively coupled to the master node edge device. The plurality of worker node edge devices defines a plurality of graphical processing units that each define a memory and a plurality of streaming multiprocessors. The orchestration system, in particular the master node edge device, can obtain a list of a plurality of virtual execution engines that each define an abstraction of the memory and a streaming multiprocessor of the plurality of streaming multiprocessors, such that each graphical processing unit of the plurality of graphical processing units defines a plurality of virtual execution engines. The system can further obtain a plurality of tasks to be performed by one or more of the plurality of graphical processing units. The system can determine an execution time associated with a first task of the plurality of tasks, and can determine task deadlines associated with each of the virtual execution engines. Based on the execution time and the task deadlines, the master node edge device can assign the first task to a first virtual execution engine of the plurality of virtualexecution engines for completion. For example, each task of the plurality of tasks can be associated with a priority, and the master node can select the first task from the plurality of tasks based on the priority of the first task.

[0031] In another example aspect, the system can retrieve a task profile associated with the first task, from a database, wherein the task profile indicates the execution time. To generate the task profile, the system can collect metrics while executing the first task on the plurality of virtual execution engines, so as to generate the task profile based on the metrics.

[0032] In some cases, before assigning the first task to the first virtual execution engine, the system can make a determination that the first virtual execution engine is available. Responsive to the determination, the system can assign the first task to the first virtual execution engine. Alternatively, the system can determine that none of the virtual execution engines in the list are available to process tasks. The system can further make a determination that the first virtual execution engine is scheduled to perform a scheduled task that has a lower priority than other scheduled tasks that the other virtual execution engines in the list are scheduled to perform. Responsive to such a determination, the system can assign the first task to the first virtual execution engine. Additionally, responsive to such a determination, the system, in particular the master node, can replace the scheduled task, from a schedule of the first virtual execution engine, with the first task. In various examples, the system can make a determination that the execution time associated with the first task of the plurality of tasks is less than the task deadline associated with the first virtual execution engine. Based on this determination, the master node edge device can assign the first task to the first virtual execution engine.

[0033] FIG. 6 illustrates an example of a computing environment within which embodiments of the present disclosure may be implemented. A computing environment 800 includes a computer system 810 that may include a communication mechanism such as a system bus 821 or other communication mechanism for communicating information within the computer system 810. The computer system 810 further includes one or more processors 820 coupled with the system bus 821 for processing the information. The industrial control network 100 may include, or be coupled to, the one or more processors 820.

[0034] The processors 820 may include one or more central processing units (CPUs), graphical processing units (GPUs), or any other processor known in the art. More generally, a processor as described herein is a device for executing machine-readable instructions stored ona computer readable medium, for performing tasks and may comprise any one or combination of, hardware and firmware. A processor may also comprise memory storing machine-readable instructions executable for performing tasks. A processor acts upon information by manipulating, analyzing, modifying, converting or transmitting information for use by an executable procedure or an information device, and / or by routing the information to an output device. A processor may use or comprise the capabilities of a computer, controller or microprocessor, for example, and be conditioned using executable instructions to perform special purpose functions not performed by a general purpose computer. A processor may include any type of suitable processing unit including, but not limited to, a central processing unit, a microprocessor, a Reduced Instruction Set Computer (RISC) microprocessor, a Complex Instruction Set Computer (CISC) microprocessor, a microcontroller, an Application Specific Integrated Circuit (ASIC), a Field-Programmable Gate Array (FPGA), a System-on-a-Chip (SoC), a digital signal processor (DSP), and so forth. Further, the processor(s) 820 may have any suitable microarchitecture design that includes any number of constituent components such as, for example, registers, multiplexers, arithmetic logic units, cache controllers for controlling read / write operations to cache memory, branch predictors, or the like. The microarchitecture design of the processor may be capable of supporting any of a variety of instruction sets. A processor may be coupled (electrically and / or as comprising executable components) with any other processor enabling interaction and / or communication there- between. A user interface processor or generator is a known element comprising electronic circuitry or software or a combination of both for generating display images or portions thereof. A user interface comprises one or more display images enabling user interaction with a processor or other device.

[0035] The system bus 821 may include at least one of a system bus, a memory bus, an address bus, or a message bus, and may permit exchange of information (e.g., data (including computer-executable code), signaling, etc.) between various components of the computer system 810. The system bus 821 may include, without limitation, a memory bus or a memory controller, a peripheral bus, an accelerated graphics port, and so forth. The system bus 821 may be associated with any suitable bus architecture including, without limitation, an Industry Standard Architecture (ISA), a Micro Channel Architecture (MCA), an Enhanced ISA (EISA), a Video Electronics Standards Association (VESA) architecture, an Accelerated Graphics Port (AGP) architecture, a Peripheral Component Interconnects (PCT) architecture, a PCI-Expressarchitecture, a Personal Computer Memory Card International Association (PCMCIA) architecture, a Universal Serial Bus (USB) architecture, and so forth.

[0036] Continuing with reference to FIG. 6, the computer system 810 may also include a system memory 830 coupled to the system bus 821 for storing information and instructions to be executed by processors 820. The system memory 830 may include computer readable storage media in the form of volatile and / or nonvolatile memory, such as read only memory (ROM) 831 and / or random access memory (RAM) 832. The RAM 832 may include other dynamic storage device(s) (e.g., dynamic RAM, static RAM, and synchronous DRAM). The ROM 831 may include other static storage device(s) (e.g., programmable ROM, erasable PROM, and electrically erasable PROM). In addition, the system memory 830 may be used for storing temporary variables or other intermediate information during the execution of instructions by the processors 820. A basic input / output system 833 (BIOS) containing the basic routines that help to transfer information between elements within computer system 810, such as during start-up, may be stored in the ROM 831. RAM 832 may contain data and / or program modules that are immediately accessible to and / or presently being operated on by the processors 820. System memory 830 may additionally include, for example, operating system 834, application programs 835, and other program modules 836. Application programs 835 may also include a user portal for development of the application program, allowing input parameters to be entered and modified as necessary.

[0037] The operating system 834 may be loaded into the memory 830 and may provide an interface between other application software executing on the computer system 810 and hardware resources of the computer system 810. More specifically, the operating system 834 may include a set of computer-executable instructions for managing hardware resources of the computer system 810 and for providing common services to other application programs (e.g., managing memory allocation among various application programs). In certain example embodiments, the operating system 834 may control execution of one or more of the program modules depicted as being stored in the data storage 840. The operating system 834 may include any operating system now known or which may be developed in the future including, but not limited to, any server operating system, any mainframe operating system, or any other proprietary or non-proprietary operating system.

[0038] The computer system 810 may also include a disk / media controller 843 coupled to the system bus 821 to control one or more storage devices for storing information andinstructions, such as a magnetic hard disk 841 and / or a removable media drive 842 (e.g., floppy disk drive, compact disc drive, tape drive, flash drive, and / or solid state drive). Storage devices 840 may be added to the computer system 810 using an appropriate device interface (e.g., a small computer system interface (SCSI), integrated device electronics (IDE), Universal Serial Bus (USB), or FireWire). Storage devices 841, 842 may be external to the computer system 810.[00391 The computer system 810 may also include a field device interface 865 coupled to the system bus 821 to control a field device 866, such as a device used in a production line. The computer system 810 may include a user input interface or GUI 861, which may comprise one or more input devices, such as a keyboard, touchscreen, tablet and / or a pointing device, for interacting with a computer user and providing information to the processors 820.

[0040] The computer system 810 may perform a portion or all of the processing steps of embodiments of the invention in response to the processors 820 executing one or more sequences of one or more instructions contained in a memory, such as the system memory 830. Such instructions may be read into the system memory 830 from another computer readable medium of storage 840, such as the magnetic hard disk 841 or the removable media drive 842. The magnetic hard disk 841 and / or removable media drive 842 may contain one or more data stores and data files used by embodiments of the present disclosure. The data store 840 may include, but are not limited to, databases (e.g., relational, object-oriented, etc.), file systems, flat files, distributed data stores in which data is stored on more than one node of a computer network, peer-to-peer network data stores, or the like. The data stores may store various types of data such as, for example, skill data, sensor data, or any other data generated in accordance with the embodiments of the disclosure. Data store contents and data files may be encrypted to improve security. The processors 820 may also be employed in a multi-processing arrangement to execute the one or more sequences of instructions contained in system memory 830. In alternative embodiments, hard-wired circuitry may be used in place of or in combination with software instructions. Thus, embodiments are not limited to any specific combination of hardware circuitry and software.

[0041] As stated above, the computer system 810 may include at least one computer readable medium or memory for holding instructions programmed according to embodiments of the invention and for containing data structures, tables, records, or other data described herein. The term “computer readable medium” as used herein refers to any medium thatparticipates in providing instructions to the processors 820 for execution. A computer readable medium may take many forms including, but not limited to, non-transitory, non-volatile media, volatile media, and transmission media. Non-limiting examples of non-volatile media include optical disks, solid state drives, magnetic disks, and magneto-optical disks, such as magnetic hard disk 841 or removable media drive 842. Non-limiting examples of volatile media include dynamic memory, such as system memory 830. Non-limiting examples of transmission media include coaxial cables, copper wire, and fiber optics, including the wires that make up the system bus 821. Transmission media may also take the form of acoustic or light waves, such as those generated during radio wave and infrared data communications.

[0042] Computer readable medium instructions for carrying out operations of the present disclosure may be assembler instructions, instruction-set-architecture (ISA) instructions, machine instructions, machine dependent instructions, microcode, firmware instructions, statesetting data, or either source code or object code written in any combination of one or more programming languages, including an object oriented programming language such as Smalltalk, C++ or the like, and conventional procedural programming languages, such as the "C" programming language or similar programming languages. The computer readable program instructions may execute entirely on the user's computer, partly on the user's computer, as a stand-alone software package, partly on the user's computer and partly on a remote computer or entirely on the remote computer or server. In the latter scenario, the remote computer may be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or the connection may be made to an external computer (for example, through the Internet using an Internet Service Provider). In some embodiments, electronic circuitry including, for example, programmable logic circuitry, field-programmable gate arrays (FPGA), or programmable logic arrays (PLA) may execute the computer readable program instructions by utilizing state information of the computer readable program instructions to personalize the electronic circuitry, in order to perform aspects of the present disclosure.

[0043] Aspects of the present disclosure are described herein with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the disclosure. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchartillustrations and / or block diagrams, may be implemented by computer readable medium instructions.

[0044] The computing environment 800 may further include the computer system 810 operating in a networked environment using logical connections to one or more remote computers, such as remote computing device 880. The network interface 870 may enable communication, for example, with other remote devices 880 or systems and / or the storage devices 841, 842 via the network 871. Remote computing device 880 may be a personal computer (laptop or desktop), a mobile device, a server, a router, a network PC, a peer device or other common network node, and typically includes many or all of the elements described above relative to computer system 810. When used in a networking environment, computer system 810 may include modem 872 for establishing communications over a network 871, such as the Internet. Modem 872 may be connected to system bus 821 via user network interface 870, or via another appropriate mechanism.

[0045] Network 871 may be any network or system generally known in the art, including the Internet, an intranet, a local area network (LAN), a wide area network (WAN), a metropolitan area network (MAN), a direct connection or series of connections, a cellular telephone network, or any other network or medium capable of facilitating communication between computer system 810 and other computers (e.g., remote computing device 880). The network 871 may be wired, wireless or a combination thereof. Wired connections may be implemented using Ethernet, Universal Serial Bus (USB), RI-6, or any other wired connection generally known in the art. Wireless connections may be implemented using Wi-Fi, WiMAX, and Bluetooth, infrared, cellular networks, satellite or any other wireless connection methodology generally known in the art. Additionally, several networks may work alone or in communication with each other to facilitate communication in the network 871.

[0046] It should be appreciated that the program modules, applications, computer-executable instructions, code, or the like depicted in FIG. 6 as being stored in the system memory 830 are merely illustrative and not exhaustive and that processing described as being supported by any particular module may alternatively be distributed across multiple modules or performed by a different module. In addition, various program module(s), script(s), plug-in(s), Application Programming Interface(s) (API(s)), or any other suitable computer-executable code hosted locally on the computer system 810, the remote device 880, and / or hosted on other computing device(s) accessible via one or more of the network(s) 871, may be provided to supportfunctionality provided by the program modules, applications, or computer-executable code depicted in the figures and / or additional or alternate functionality. Further, functionality may be modularized differently such that processing described as being supported collectively by the collection of program modules depicted in the figures may be performed by a fewer or greater number of modules, or functionality described as being supported by any particular module may be supported, at least in part, by another module. In addition, program modules that support the functionality described herein may form part of one or more applications executable across any number of systems or devices in accordance with any suitable computing model such as, for example, a client-server model, a peer-to-peer model, and so forth. In addition, any of the functionality described as being supported by any of the program modules depicted in the figures may be implemented, at least partially, in hardware and / or firmware across any number of devices.

[0047] It should further be appreciated that the computer system 810 may include alternate and / or additional hardware, software, or firmware components beyond those described or depicted without departing from the scope of the disclosure. More particularly, it should be appreciated that software, firmware, or hardware components depicted as forming part of the computer system 810 are merely illustrative and that some components may not be present or additional components may be provided in various embodiments. While various illustrative program modules have been depicted and described as software modules stored in system memory 530, it should be appreciated that functionality described as being supported by the program modules may be enabled by any combination of hardware, software, and / or firmware. It should further be appreciated that each of the above-mentioned modules may, in various embodiments, represent a logical partitioning of supported functionality. This logical partitioning is depicted for ease of explanation of the functionality and may not be representative of the structure of software, hardware, and / or firmware for implementing the functionality. Accordingly, it should be appreciated that functionality described as being provided by a particular module may, in various embodiments, be provided at least in part by one or more other modules. Further, one or more depicted modules may not be present in certain embodiments, while in other embodiments, additional modules not depicted may be present and may support at least a portion of the described functionality and / or additional functionality. Moreover, while certain modules may be depicted and described as sub-modulesof another module, in certain embodiments, such modules may be provided as independent modules or as sub-modules of other modules.

[0048] Although specific embodiments of the disclosure have been described, one of ordinary skill in the art will recognize that numerous other modifications and alternative embodiments are within the scope of the disclosure. For example, any of the functionality and / or processing capabilities described with respect to a particular device or component may be performed by any other device or component. Further, while various illustrative implementations and architectures have been described in accordance with embodiments of the disclosure, one of ordinary skill in the art will appreciate that numerous other modifications to the illustrative implementations and architectures described herein are also within the scope of this disclosure. In addition, it should be appreciated that any operation, element, component, data, or the like described herein as being based on another operation, element, component, data, or the like can be additionally based on one or more other operations, elements, components, data, or the like. Accordingly, the phrase “based on,” or variants thereof, should be interpreted as “based at least in part on.”

[0049] Although embodiments have been described in language specific to structural features and / or methodological acts, it is to be understood that the disclosure is not necessarily limited to the specific features or acts described. Rather, the specific features and acts are disclosed as illustrative forms of implementing the embodiments. Conditional language, such as, among others, “can,” “could,” “might,” or “may,” unless specifically stated otherwise, or otherwise understood within the context as used, is generally intended to convey that certain embodiments could include, while other embodiments do not include, certain features, elements, and / or steps. Thus, such conditional language is not generally intended to imply that features, elements, and / or steps are in any way required for one or more embodiments or that one or more embodiments necessarily include logic for deciding, with or without user input or prompting, whether these features, elements, and / or steps are included or are to be performed in any particular embodiment.

[0050] The flowchart and block diagrams in the Figures illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of the present disclosure. In this regard, each block in the flowchart or block diagrams may represent a module, segment, or portion of instructions, which comprises one or more executable instructions for implementing thespecified logical function(s). In some alternative implementations, the functions noted in the block may occur out of the order noted in the Figures. For example, two blocks shown in succession may, in fact, be executed substantially concurrently, or the blocks may sometimes be executed in the reverse order, depending upon the functionality involved. It will also be noted that each block of the block diagrams and / or flowchart illustration, and combinations of blocks in the block diagrams and / or flowchart illustration, can be implemented by special purpose hardware-based systems that perform the specified functions or acts or carry out combinations of special purpose hardware and computer instructions.

Claims

CLAIMSWhat is claimed is:

1. A method performed within an industrial network that defines a master node edge device and a plurality of worker node edge devices communicatively coupled to the master node edge device, the plurality of worker node edge devices defining a plurality of graphical processing units that each define a memory and a plurality of streaming multiprocessors, the method comprising: obtaining a list of a plurality of virtual execution engines that each define an abstraction of the memory and a streaming multiprocessor of the plurality of streaming multiprocessors, such that each graphical processing unit of the plurality of graphical processing units defines a plurality of virtual execution engines; obtaining a plurality of tasks to be performed by one or more of the plurality of graphical processing units; determining an execution time associated with a first task of the plurality of tasks; determining task deadlines associated with each of the virtual execution engines; and based on the execution time and the task deadlines, the master node edge device assigning the first task to a first virtual execution engine of the plurality of virtual execution engines for completion.

2. The method as recited in claim 1, wherein each task of the plurality of tasks is associated with a priority, the method further comprising: selecting the first task from the plurality of tasks based on the priority of the first task.

3. The method as recited in claim 1, wherein determining the execution time of the first task comprises: retrieving a task profile associated with the first task, from a database, wherein the task profile indicates the execution time.

4. The method as recited in claim 3, the method further comprising:collecting metrics while executing the first task on the plurality of virtual execution engines; and generating the task profile based on the metrics.

5. The method as recited in claim 1, the method further comprising: before assigning the first task to the first virtual execution engine, making a determination that the first virtual execution engine is available; and responsive to the determination, assigning the first task to the first virtual execution engine.

6. The method as recited in claim 1, the method further comprising: determining that none of the virtual execution engines in the list are available to process tasks; making a determination that the first virtual execution engine is scheduled to perform a scheduled task that has a lower priority than other scheduled tasks that the other virtual execution engines in the list are scheduled to perform; and responsive to the determination, assigning the first task to the first virtual execution engine.

7. The method as recited in claim 6, the method further comprising: responsive to the determination, replacing the scheduled task, from a schedule of the first virtual execution engine, with the first task.

8. The method as recited in claim 1, the method further comprising: making a determination that the execution time associated with the first task of the plurality of tasks is less than the task deadline associated with the first virtual execution engine; and based on determination, the master node edge device assigning the first task to the first virtual execution engine.

9. An orchestration system comprising: a master node edge device and a plurality of worker node edge devices communicatively coupled to the master node edge device, the plurality of worker node edge devices defining aplurality of graphical processing units that each define a memory and a plurality of streaming multiprocessors, wherein the master node edge device is configured to: obtain a list of a plurality of virtual execution engines that each define an abstraction of the memory and a streaming multiprocessor of the plurality of streaming multiprocessors, such that each graphical processing unit of the plurality of graphical processing units defines a plurality of virtual execution engines; obtain a plurality of tasks to be performed by one or more of the plurality of graphical processing units; determine an execution time associated with a first task of the plurality of tasks; determine task deadlines associated with each of the virtual execution engines; and based on the execution time and the task deadlines, assign the first task to a first virtual execution engine of the plurality of virtual execution engines for completion.

10. The orchestration system as recited in claim 9, wherein each task of the plurality of tasks is associated with a priority, the master node edge device further configured to: select the first task from the plurality of tasks based on the priority of the first task.

11. The orchestration system as recited in claim 9, the master node edge device further configured to: retrieve a task profile associated with the first task, from a database, wherein the task profile indicates the execution time.

12. The orchestration system as recited in claim 11, the master node edge device further configured to: collect metrics while executing the first task on the plurality of virtual execution engines; and generate the task profile based on the metrics.

13. The orchestration system as recited in claim 9, the master node edge device further configured to: before assigning the first task to the first virtual execution engine, make a determination that the first virtual execution engine is available; and responsive to the determination, assign the first task to the first virtual execution engine.

14. The orchestration system as recited in claim 9, the master node edge device further configured to: determine that none of the virtual execution engines in the list are available to process tasks; make a determination that the first virtual execution engine is scheduled to perform a scheduled task that has a lower priority than other scheduled tasks that the other virtual execution engines in the list are scheduled to perform; and responsive to the determination, assign the first task to the first virtual execution engine.

15. The orchestration system as recited in claim 14, the master node edge device further configured to: responsive to the determination, replace the scheduled task, from a schedule of the first virtual execution engine, with the first task.

Images