GPU Receive for PME/PP GPU Force Communications
commit5b594f3bc695fe439c4dcbedc09b38c0fe3cc42f
authorAlan Gray <alang@nvidia.com>
Sat, 31 Aug 2019 18:20:52 +0000 (31 11:20 -0700)
committerSzilárd Páll <pall.szilard@gmail.com>
Fri, 18 Oct 2019 22:52:11 +0000 (19 00:52 +0200)
tree364f4af135a2ef714590897351ce9f94c5ce84be
parent6a3add7be671a7c57805ca9070ac66b2f714d881
GPU Receive for PME/PP GPU Force Communications

This change extends the PME/PP GPU force communication functionality
to allow the force buffer to be recieved direct to GPU memory on the
PP task.

Implements part of #2817
Refs #3158 #3159

Change-Id: I5b1cff1846c7c3bd966b6bf9c0af72769600ef18
src/gromacs/ewald/pme.h
src/gromacs/ewald/pme_pp.cpp
src/gromacs/ewald/pme_pp_comm_gpu.h
src/gromacs/ewald/pme_pp_comm_gpu_impl.cpp
src/gromacs/ewald/pme_pp_comm_gpu_impl.cu
src/gromacs/ewald/pme_pp_comm_gpu_impl.h
src/gromacs/mdlib/sim_util.cpp