Bump rocm-systems from c76140f to 8017ffe

[systems-assistant]

Bumps ROCm/rocm-systems from c76140f to 8017ffe.

Commits

See full comparison here:

ROCm/rocm-systems@c76140f...8017ffe

[araravik-psd]

Sanity check issues for Windows seem to be runner issues, both gfx110x and gfx1151 seem to be missing gpu nodes or are not recognizing them.

hipGetDeviceCount: Returned hipErrorNoDevice
Failed to get device count: no ROCm-capable device is detected

[ScottTodd]

Sanity check issues for Windows seem to be runner issues, both gfx110x and gfx1151 seem to be missing gpu nodes or are not recognizing them.

hipGetDeviceCount: Returned hipErrorNoDevice Failed to get device count: no ROCm-capable device is detected

Agreed these look like machine issues. I can't reproduce locally on Windows gfx1100. Using https://github.com/ROCm/TheRock/actions/runs/24282325338/job/71112020580?pr=4474,

python build_tools/artifact_manager.py fetch --run-id=24282325338 --stage=all --flatten --amdgpu-families=gfx110X-all --output-dir=D:/scratch/therock/24282325338/

cd D:/scratch/therock/24282325338/bin
.\hipInfo.exe
# valid output

cd D:/scratch/therock/24282325338/lib/llvm/bin
.\offload-arch.exe
# gfx1100
# gfx1100

[jayhawk-commits]

Windows gfx110X test results look promising

• hip-tests on ROCR xfail, should not block
• if hipsparse passes on retry, then I will go ahead and merge this PR

[jayhawk-commits]

Windows gfx110X hipsparse issue occurs on retries
Issue did not occur on the rocm-libraries update PR #4474

[araravik-psd]

Windows gfx110X hipsparse issue occurs on retries Issue did not occur on the rocm-libraries update PR #4474

hipsparse failures are seen on #4473
#4471 too, following up with devs and will file an issue for this

[araravik-psd]

Issues for hipsparse are being looked at: #4533. ROCR hip-tests failures are expected.

Bypassing and merging this to move rocm-systems ahead in therock.