Method for sharing stream memory of heterogeneous multi-processor

Abstract

The invention provides a method for sharing the stream memory of a heterogeneous multi-processor. The method comprises the following steps: an application program runs on a master processor and an API is called for the first time, and one or more executable programs are encoded from the source code containing local variables for a plurality of processor units with stream memory; and then the API is called for the second time so as to load one or more executable programs to a plurality of processor units; collateral execution is conducted on a plurality of treads; when in loading, local storage units are allocated from the local storage of a processor; in addition, when in loading, a first stream storage unit is allocated from the stream memory; when a processing unit executes a plurality of treads simultaneously, the threads access to the values of the variables on the basis of the storage units of the stream memory; the source program containing stream variables further comprises the following steps: the API is called for the third time; in the stream memory, a second stream storage unit is allocated for the stream variables; based on the second stream storage unit, the variable values of the stream variables are accessed through a plurality of processor units.

Description

technical field [0001] The invention relates to a data parallel computing technology, in particular to a method for sharing stream memory when performing data parallel computing through heterogeneous multi-processor CPUs and GPUs. Background technique [0002] As GPUs are gradually incorporated into high-performance parallel computing devices, GPUs are regarded as general-purpose computing devices, and more and more applications are developed to complete data parallel computing. Today, we design these applications with professional interfaces and professional GPU devices provided by suppliers, so that even if the CPU and GPU are used together in the data processing system, the CPU will not be overloaded and the application can run on different vendors' GPUs. on the GPU. [0003] However, as more and more CPUs are embedded into multi-cores to perform data parallel computing, more and more data processing tasks can be completed with CPUs and GPUs. Processors combined with mu...

AI Technical Summary

Problems solved by technology

A combination of multiple CPUs or GPUs is abbreviated as CPUs and GPUs. Traditionally, GPUs and CPUs are compiled through different program environments, so the mutual adaptability of CPUs and GPUs is not very good.

Images