tree ce0d280aa4b6d4f275730ef3313cad6b3b113aec
parent 0d3452218a8029a6bfb5fcc2af8f8cb5c212080c
author Jan Patrick Lehr <jplehr@users.noreply.github.com> 1699993804 +0100
committer Copybara-Service <copybara-worker@google.com> 1699994166 -0800

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.

GitOrigin-RevId: 5c22b907dc557d739d5141fad04acac9c97c09b9