| DOI | |
| Creators (name, orcid, affilation) |
Goz David, ORCID: 0000-0001-9808-2283, INAF-OATs |
| Main Affiliation | OATs |
| Titles | Performance of direct N-body code on ARM64 SoC |
| Series | INAF-OATs Technical Report |
| Series Number | 241 |
| Publication Year | 2018 |
| Resource type | text |
| Language | en |
| abstract | The ExaNeSt H2020 project will produce a prototype based on ARM64 CPUs and FPGA-based accelerators implementing a co-design approach where scientific applications requirements are driving the hardware design. We present strategies adopted in order to port our direct N-body code on an ARM SoC cluster, called INCAS, deployed at INAF-OATs. Each computational node is based on 64-bit ARMv8 Cortex-A72/Cortex-A53 core design, powered by the Mali-T864 GPU. We present the strategies to optimize the code and we discuss in details results of performance tests carried on ARM CPUs and GPUs, showing that embedded GPUs can be effectivelyused to accelerate the calculation, and that are promising also because of their energy efficiency, which is an important design in future exascale platforms. |