[all-commits] [llvm/llvm-project] 5c22b9: Reland [OpenMP][libomptarget] Enable parallel copi...
Jan Patrick Lehr via All-commits
all-commits at lists.llvm.org
Tue Nov 14 12:30:18 PST 2023
Branch: refs/heads/main
Home: https://github.com/llvm/llvm-project
Commit: 5c22b907dc557d739d5141fad04acac9c97c09b9
https://github.com/llvm/llvm-project/commit/5c22b907dc557d739d5141fad04acac9c97c09b9
Author: Jan Patrick Lehr <jplehr at users.noreply.github.com>
Date: 2023-11-14 (Tue, 14 Nov 2023)
Changed paths:
M openmp/libomptarget/plugins-nextgen/amdgpu/src/rtl.cpp
Log Message:
-----------
Reland [OpenMP][libomptarget] Enable parallel copies via multiple SDM… (#72307)
…A engines (#71801)
This enables the AMDGPU plugin to use a new ROCm 5.7 interface to
dispatch asynchronous data transfers across SDMA engines.
The default functionality stays unchanged, meaning that all data
transfers are enqueued into a H2D queue or an D2H queue, depending on
transfer direction, via the HSA interface used previously.
The new interface can be enabled via the environment variable
`LIBOMPTARGET_AMDGPU_USE_MULTIPLE_SDMA_ENGINES=true` when libomptarget
is built against a recent ROCm version (5.7 and later). As of now,
requests are distributed in a round-robin fashion across available SDMA
engines.
More information about the All-commits
mailing list