Embedded high-throughput computing system

A computing system and high-throughput technology, applied in the direction of computing, program control design, multi-program device, etc., can solve the problem of low resource utilization, reduce the cost of the entire life cycle, support intelligent task processing capabilities, and improve resources The effect of utilization

Pending Publication Date: 2022-05-06
XIAN AVIATION COMPUTING TECH RES INST OF AVIATION IND CORP OF CHINA
0 Cites 0 Cited by

AI-Extracted Technical Summary

Problems solved by technology

[0004] The purpose of the present invention is to provide an embedded high-throughput computing ...
View more

Method used

Task receiver is also used for carrying out task monitoring to task management system, determines whether current task is new task, if not, current task is sent to resource manager, as not, current task is sent to resource manager and feed back the first signal, the resource manager invokes the historical data according to the first signal, the historical data includes resources allocated for processing the current task, and can realize fast processing of the task, and/or,
With reference test application as input, generate the embedded high-throughput computing system intelligent resource allocation software that satisfies performance index requiremen...
View more

Abstract

The invention discloses an embedded high-throughput computing system, belongs to the technical field of embedded computing systems, and aims to meet the requirements of intelligent task scheduling and resource management in the embedded high-throughput computing system, the high-throughput data processing is taken as the center by complex embedded computing systems such as airborne computing systems in the future. In combination with artificial intelligence algorithms such as deep reinforcement learning and the like, the intelligent resource management architecture of the embedded high-throughput computing system is provided and comprises a task receiver, a resource pool monitor, an intelligent resource distributor and a resource manager. The intelligent resource management architecture of the embedded high-throughput computing system and the verification method can improve the intelligent task processing capability and the resource utilization rate of the airborne embedded computing system.

Application Domain

Program initiation/switchingResource allocation +1

Technology Topic

Embedded systemHigh-throughput computing +7

Image

  • Embedded high-throughput computing system

Examples

  • Experimental program(1)

Example Embodiment

[0014] The following describes the embodiment of the invention through specific specific examples. Those skilled in the art can easily understand other advantages and functions of the invention from the contents disclosed in the specification. Obviously, the described embodiments are only part of the embodiments of the invention, not all of them. The invention can also be implemented or applied through other different specific embodiments, and various details in the specification can also be modified or changed based on different viewpoints and applications without departing from the spirit of the invention. It should be noted that the following embodiments and the features in the embodiments can be combined with each other without conflict. Based on the embodiments of the invention, all other embodiments obtained by ordinary technicians in the art without creative work belong to the protection scope of the invention.
[0015] It should be noted that various aspects of the embodiments within the scope of the appended claims are described below. It should be apparent that the aspects described herein can be embodied in a wide variety of forms, and any specific structure and / or function described herein is illustrative only. Based on the invention, those skilled in the art should understand that one aspect described herein can be implemented independently of any other aspect, and two or more of these aspects can be combined in various ways. For example, any number of aspects set forth herein may be used to implement equipment and / or practice methods. In addition, the apparatus and / or the method may be implemented using structures and / or functionality other than one or more of the aspects set forth herein.
[0016] It should also be noted that the diagrams provided in the following embodiments only illustrate the basic concept of the invention in a schematic way. The diagrams only show the components related to the invention rather than drawing according to the number, shape and size of components in the actual implementation. The type, number and proportion of components in the actual implementation can be changed at will, and the component layout type may be more complex.
[0017] In addition, in the following description, specific details are provided to facilitate a thorough understanding of the example. However, those skilled in the art will understand that aspects can be practiced without these specific details. In order to enable those in the technical field to better understand the scheme of the invention, the invention is further described in detail below in combination with the accompanying drawings and specific embodiments. The terms "first" and "second" are only used for descriptive purposes and cannot be understood as indicating or implying relative importance or implicitly indicating the number of indicated technical features. Thus, the features defined with "first" and "second" can explicitly or implicitly include one or more of the features. In the description of the invention, unless otherwise specified, "multiple" means two or more.
[0018] as Figure 1 The embedded high-throughput computing system is applicable to the data interaction of task management system under high-throughput. The system includes:
[0019] Task receiver: receiving the tasks sent by the task management system and sending data to the task management system;
[0020] Resource pool detector: including general computing module and multiple independent function processing modules. The general computing module includes multiple CPU processing units. Each CPU processing unit can process comprehensive data, and the independent function processing module can process single data;
[0021] The resource manager can monitor the working status of the resource pool detector and the task receiver, call the resource pool detector resources to process the tasks, and input and output the results;
[0022] Intelligent resource distributor: obtain the result output and fuse it. The fused result output is sent to the task receiver, and the task receiver feeds back the fused result output to the task management system. Fusion, for example, the fusion of sound, image, size, etc. to form the overall data, mainly for the data processed by multiple independent function processing modules.
[0023]Taking the benchmark application as the input, the intelligent resource allocation software of the embedded high-throughput computing system that meets the performance index requirements is generated, which improves the resource utilization of the embedded high-throughput computing system, reduces the whole life cycle cost of the system, supports the intelligent task processing ability, and realizes the reasonable allocation of resources.
[0024] Specifically, the resource pool detector is used for priority planning of the calculation module and multiple independent function processing modules, and the calculation module has the highest priority, and multiple independent function processing modules are sorted according to priority, so as to avoid congestion in the process of data interaction;
[0025] The resource manager includes a monitoring module, which is used for real-time state monitoring of the resource pool detector to determine whether the computing module and independent function processing module are idle or busy;
[0026] The task receiver is also used to monitor the task of the task management system to determine whether the current task is a new task. If not, send the current task to the resource manager. If not, send the current task to the resource manager and feed back the first signal. The resource manager calls the historical data according to the first signal. The historical data includes the resources allocated by the current task, which can realize the rapid processing of the task, and / or,
[0027] Preset the task processing cycle to judge whether the current task is completed within the preset cycle. If yes, do not obtain the current task signal. If no, obtain the current task signal processed by the task management system and feed back the signal to stop processing to the task management system, indicating that the system configuration of the task management system is low and the processing time is long, and / or,
[0028] Judge whether the task management system fails when processing the current task. If yes, obtain the task currently processed by the task management system. If no, do not obtain the task currently processed by the task management system. The purpose is to realize the monitoring of the task management system. When the task cannot be handled, or the processing is slow or there is a system failure, the task will be transferred to processing.
[0029] As the specific embodiment provided in this case, the data of the calculation module of the resource pool detector is multiple, and the priority is the same. The CPU processing unit can adopt the products of the prior art. If the calculated output data does not need to be fused, it can be sent directly. If fusion is needed, it can be fused through the intelligent resource distributor. Specifically:
[0030] It includes a preset number of independent function processing modules with different processing functions. There are multiple independent function processing modules with the same priority.
[0031] As the specific embodiment provided in this case, the resource pool detector also includes a decomposition module, wherein:
[0032] The decomposition module is used to receive the task information of the task receiver and judge whether the task information is a single resource calculation. If yes, judge whether there are idle CPU processing units in the calculation module. If yes, give priority to randomly allocating idle CPU processing units for processing. If no, match the corresponding independent function processing module for processing;
[0033] If not, divide the task information into multiple subtask information and judge whether there are idle CPU processing units in the calculation module. If yes, give priority to randomly assigning idle CPU processing units for processing. If not, obtain and determine the address of the processing independent function processing module corresponding to all subtask information and sort according to priority, that is, the first sort, According to the first order, all subtask information is allocated to the corresponding and idle independent function processing module for processing.
[0034] The information of multiple subtasks is processed in parallel to shorten the time-consuming of calculation. The subtasks can be processed without queuing.
[0035] As the specific embodiment provided in this case, the resource pool detector also includes a monitoring unit, which is used to monitor whether the CPU processing unit or independent function processing module fails when the subtask information or task information is processed. If so, call similar resources to process the subtask information or task information. If not, do not call. The purpose is to conduct internal monitoring to avoid outputting invalid data.
[0036] Further, a plurality of independent functional processing modules at least include a signal processing unit, an image processing unit, an intelligent computing unit and a data storage unit to meet the separate processing of different tasks and improve the calculation efficiency.
[0037] In the above scheme, the intelligent resource distributor at least includes heuristic algorithm module, gas gathering learning algorithm module, graph neural network module and deep learning network module, wherein:
[0038] At least one model is called to fuse the information processing results of multiple subtasks, and the fused processing data is fed back to the task receiver.
[0039] The products provided by the invention are described in detail above. In this paper, a specific example is applied to explain the principle and embodiment of the invention. The description of the above embodiment is only used to help understand the core idea of the invention. It should be noted that ordinary technicians in the technical field can also make several improvements and modifications to the invention without departing from the principle of invention and creation, and these improvements and modifications also fall into the protection scope of the invention claims.

PUM

no PUM

Description & Claims & Application Information

We can also present the details of the Description, Claims and Application information to help users get a comprehensive understanding of the technical details of the patent, such as background art, summary of invention, brief description of drawings, description of embodiments, and other original content. On the other hand, users can also determine the specific scope of protection of the technology through the list of claims; as well as understand the changes in the life cycle of the technology with the presentation of the patent timeline. Login to view more.

Similar technology patents

Method for combining resource allocation and content caching in F-RAN architecture

ActiveCN109951849AReduce the pressure on the fronthaul linkImprove resource utilization
Owner:CHONGQING UNIV OF POSTS & TELECOMM

Classification and recommendation of technical efficacy words

  • Improve resource utilization

Statistical-prediction-based automatic cloud CDN (Content Delivery Network) resource automatic deployment method

InactiveCN102801792AImprove resource utilizationReduce energy consumption and operation and maintenance costs
Owner:SOUTH CHINA UNIV OF TECH

Dual-card dual-standby terminal and data communication method

InactiveCN105101164AImprove resource utilizationAvoid data service call drop
Owner:NUBIA TECHNOLOGY CO LTD
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Try Eureka
PatSnap group products