Draft: GPU aware RMA #3875

sb17v · 2019-06-25T20:18:56Z

Pull Request Description

Basic implementation of GPU aware MPI_Accumulate operation without GPU Direct RDMA Support

Reference appropriate issues (with "Fixes" or "See" as appropriate)
Passes tests (included warning check)
Confirm whitespace/style checkers are happy (or has a good reason for being bad)
Commits are self-contained and do not do two things at once
Remove xfail from the test suite when fixing a test
Commit message is of the form: module: short description and follows good practice
Add comments such that someone without knowledge of the code could understand
Add Devel Docs in the doc/ directory for any new code design

…UDirect RDMA (Buffer copy D-H and H-D)

sb17v force-pushed the GPU_aware_RMA branch 2 times, most recently from cf9a314 to a04af5e Compare June 26, 2019 15:47

sb17v added 4 commits June 26, 2019 14:22

Adding CUDA detection to configure

e933339

Adding support for GPU Buffer transfer using AM path without using GP…

42c1f0a

…UDirect RDMA (Buffer copy D-H and H-D)

Adding support for compuatation offloading to GPU

eeace26

Adding computation offloading support for different OP and Datatypes

4eea475

gcongiu force-pushed the GPU_aware_RMA branch from a04af5e to 4eea475 Compare June 26, 2019 19:23

gcongiu changed the title ~~GPU aware RMA~~ Draft: GPU aware RMA Jun 27, 2019

CUDA operation profiling and code cleanup

d4ef670

hzhou added this to GPU in Threading and GPU Hackathon Dec 7, 2019

pavanbalaji closed this May 14, 2020