A method, device and medium for determining the multiplication speed of concurrency

Abstract

The embodiment of the invention discloses a concurrent quantity improvement speed multiplication determination method and device and a medium. The method comprises the steps: hardware parameter information is acquired, wherein the hardware parameter information comprises parameter information of a stand-alone hardware environment or parameter information of a distributed hardware environment; concurrent quantity improvement speed multiplication is determined according to the hardware parameter information; and the operation position of data calculation is determined according to the concurrentquantity improvement speed multiplication. According to the embodiment, hardware parameter information is obtained, the concurrent quantity improvement speed multiplication is estimated and determined before data calculation operation according to the acquired hardware parameter information, and the actual operation position of data calculation is determined according to the estimated concurrentquantity improvement speed multiplication, so that the concurrent quantity improvement effect is improved, unnecessary work expenditure is reduced, and the user experience effect is improved.

Description

technical field [0001] The embodiments of the present invention relate to the field of high-performance computing, and in particular, to a method, device, and medium for determining the multiplied speed of concurrency increase. Background technique [0002] With the continuous enrichment of hardware functions and the gradual maturity of software development environment, Graphics Processing Unit (GPU) has begun to be used in the field of general computing. In order to pursue high performance, GPU often contains hundreds of core computing units and high-level Density of computing resources. In a stream processing system, the CPU algorithm processes a steady stream of stream data, and the coprocessor can accelerate traditional CPU-based computing programs, thereby effectively improving the concurrent processing capability of the stream processing system. [0003] In the process of implementing the present invention, the inventor found that the existing technology has the follo...

AI Technical Summary

Problems solved by technology

[0003] In the process of implementing the present invention, the inventor found that the existing technology has the following defects: when determining the concurrency improvement effect of GPU on CPU in a heterogeneous streaming computing environment, evaluation and data program migration based on manual experience are usually used The actual calculation results are used to determine the effect of concurrency improvement. However, the use of manual experience for evaluation usually results in inaccurate actual evaluation results due to lack of strong evidence. Therefore, the data calculation is migrated based on the evaluation results to determine the data calculation After running the location, there are often cases where the concurrency improvement effect is not ideal, which increases a lot of unnecessary work overhead and affects the user experience effect

Images