Accel-OpenMP-Portability-Scalability

Over the last decade, most of the increase in computing power has been gained by advances in accelerated many-core architec- tures, mainly in the form of GPGPUs. While accelerators achieve phe- nomenal performances in various computing tasks, their utilization re- quires code adaptations and transformations. Thus, OpenMP, the most common standard for multi-threading in scientific computing applica- tions, introduced offloading capabilities between host (CPUs) and accel- erators since v4.0, with increasing support in the successive v4.5, v5.0, v5.1, and the latest v5.2 versions. Recently, two state-of-the-art GPUs – the Intel Ponte Vecchio Max 1100 and the NVIDIA A100 GPUs – were released to the market, with the oneAPI and GNU LLVM-backed com- pilation for offloading, correspondingly. In this work, we present early performance results of OpenMP offloading capabilities to these devices while specifically analyzing the portability of advanced directives (us- ing SOLLVE’s OMPVV test suite) and the scalability of the hardware in representative scientific mini-app (the LULESH benchmark). Our results show that the vast majority of the offloading directives in v4.5 and 5.0 are supported in the latest oneAPI and GNU compilers; however, the support in v5.1 and v5.2 is still lacking. From the performance perspec- tive, we found that the PVC1100 and A100 are relatively comparable on the LULESH benchmark. While the A100 is slightly better due to faster memory bandwidth, the PVC1100 reaches the next problem size (400^3) scalably due to the larger memory size.