Published July 14, 2018 | Version v1

Low latency RDMA for High Performance Computing on ARM platforms

Description

In HPC, low latency communication between remote processes is crucial to application performance. InfiniBand networks can reduce the latency but require special and costly network interface cards, which are loosely coupled with CPU, thus needing to constantly re-pin pages. In this work we leverage the SMMU available in the Zynq Ultrascale+ MPSoC’s in order to perform user-level initiated zero-copy transfers in a cluster of 64-bit ARM-based nodes, achieving a 5x latency improvement over kernel-based solutions.

Files

Acaces2017_FORTH_4.pdf

Files (306.4 kB)

Name Size Download all
md5:396cc5a91e4f421b8a2a2fda49eb18cb
306.4 kB Preview Download

Additional details

Funding

European Commission
ExaNeSt - European Exascale System Interconnect and Storage 671553