Terminal computing power deployment system and method for high-performance elastic computing HEC

Abstract

The invention provides a terminal computing power deployment system and method for high-performance elastic computing HEC. The system comprises a receiving device which is used for collecting terminalinformation; A processing device WHICH comprises an HEC main chip system and an HEC sub-chip array, WHREIN the HEC main chip system is used for configuring at least one HEC sub-chip in the HEC sub-chip array to process terminal information according to the terminal information after acquiring the terminal information; at least one HEC sub-chip which is used for generating a calculation result according to the terminal information and sending the calculation result to the HEC main chip system; and a driving device which is used for receiving the calculation result and driving the terminal to execute motion according to the calculation result. According to the HEC main chip system provided by the invention, the HEC protocol controller and the bridging module are used for integrating different numbers of HEC sub-chips into an integral computing system; a computing resource pool is formed by computing resources in all the HEC sub-chips, and the HEC main chip system schedules the HEC sub-chips in a unified mode, so that on-demand flexible deployment of system computing power under a unified software computing architecture and a programming interface is achieved.

Description

technical field [0001] The present application relates to the technical field of high-performance elastic computing, and in particular to a terminal computing power deployment system and method for high-performance elastic computing (HEC). Background technique [0002] At present, with the popularization of intelligence, there are more and more developments and applications of intelligent terminals such as robots and self-driving cars. In the actual application process of these smart devices, they need to calculate according to the external environment or instructions, and then intelligently execute the corresponding commands. However, after the computing platform of these terminals is selected, its maximum computing power is fixed and cannot be modified. If the computing power of the terminal exceeds the computing platform, the entire computing platform needs to be replaced, which brings great trouble. In addition, if multiple computing platforms are cascaded or paralleliz...

AI Technical Summary

Problems solved by technology

However, after the computing platform of these terminals is selected, its maximum computing power is fixed and cannot be modified. If the computing power of the terminal exceeds the computing platform, the entire computing platform needs to be replaced, which brings great trouble

In addition, if multiple computing platforms are used for cascade or parallel computing, their software computing architecture and programming interface are not unified, which will make it difficult or even impossible to deploy software and algorithms.

Images