A First Step Toward Support for MPI Partitioned Communication on SYCL-Programmed FPGAs
DescriptionVersion 4.0 of the Message Passing Interface standard introduced the concept of Partitioned Communication, which adds support for multiple contributions to a communication buffer. Although initially targeted at multithreaded MPI applications, Partitioned Communication currently receives attraction in the context of accelerators, especially GPUs. In this publication, it is demonstrated that this communication concept can be implemented for SYCL-programmed FPGAs. This includes a discussion of the design space and the presentation of a prototype implementation. Experimental results show that a lightweight implementation on top of an existing MPI library is possible. The presented approach also reveals issues in both the SYCL and the MPI standard, which needs to be addressed for improved support for the intended communication style.
Event Type
Workshop
TimeMonday, 14 November 202210:30am - 11am CST
LocationC146
Registration Categories
W
Session Formats
Recorded
Back To Top Button