OpenACC Execution Models for Manycore Processor with ARM SVE
Mitsuhisa Sato, Miwako Tsuji · 2023
OpenACC is designed to offer performance portability across CPUs with SIMD extensions and accelerators based on GPU or manycore architecture. We are working on the design of OpenACC compiler for A64FX manycore processor with Arm SVE. We use a source-to-source compiler infrastructure, Omni compiler, to generate the code with OpenMP SIMD directive and generate a translated code in OpenCL which is executed by OpenCL runtime for CPUs. We expect OpenACC to be used to exploit different levels of parallelism for NUMA node, cores in a NUMA node and SIMD vector by Arm SVE in A64FX manycore processor, more than conventional flat OpenMP parallel loops. The preliminary results of our source-to-source transformation with OpenMP SIMD directive is presented.