Add CI tests for AMD GPU accelerator detection (real-hardware fixtures)

[hmeiland]

Summary

Adds workflow-based CI tests for AMD GPU accelerator detection in eessi_archdetect.sh, building on the detection logic proposed in #205. All tests use output captured from real hardware (an AMD Radeon Pro V710 / NAVI32, gfx1101), so no AMD GPU is required in CI.

This addresses the AMD-CI gap tracked in #214, complementing the detection work in #205.

What's covered

accelpath()'s AMD path (from #205) has two detection methods; this PR tests both:

• Method 1 — KFD sysfs (/sys/devices/virtual/kfd/kfd/topology/nodes/*/properties, no amd-smi/Python required). This is the path that fires on real AMD hardware where amdgpu-arch isn't in PATH. To make it testable without a GPU, the topology root is made overridable via $EESSI_KFD_TOPOLOGY_ROOT (defaults to the real sysfs path — behavior is unchanged when unset). Fixtures are captured nodes/*/properties files:
  • v710 → node 0 is a CPU node (gfx_target_version 0, must be skipped), node 1 is the GPU (gfx_target_version 110001 → gfx1101). Verifies the gfx_target_version → gfxMAJMINSTEP decode and the CPU-node skip.
  • cpu_only → only a CPU node → no accelerator → exit 2.
• Method 2 — amd-smi fallback: a fake amd-smi on $PATH replays real amd-smi static --asic output (TARGET_GRAPHICS_VERSION: gfx1101), mirroring the existing nvidia-smi test pattern.

Changes

• init/eessi_archdetect.sh: one-line change making the KFD topology root overridable via $EESSI_KFD_TOPOLOGY_ROOT (test seam; default unchanged).
• .github/workflows/tests_archdetect_amd_gpu.yml: new workflow, analogous to tests_archdetect_nvidia_gpu.yml, with kfd_sysfs (v710, cpu_only) and amd_smi (v710) matrices.
• tests/archdetect/kfd/{v710,cpu_only}/: captured KFD topology fixtures + expected output.
• tests/archdetect/amd-smi/: amd-smi stub + expected output.

Verification

Verified locally against the #205 branch, and the gfx1101 result was confirmed on the live V710 (both KFD gfx_target_version=110001 and amd-smi TARGET_GRAPHICS_VERSION: gfx1101 decode to accel/amd/gfx1101):

Scenario	Result
KFD / V710 fixture	accel/amd/gfx1101
KFD / cpu_only	exit 2 (CPU node skipped)
amd-smi stub fallback	accel/amd/gfx1101
NVIDIA cc80 stub (regression)	accel/nvidia/cc80 (unchanged)
No accelerators	exit 2 (unchanged)

Notes

• This depends on the AMD detection logic in Add AMD GPU support to accelpath (WIP) #205; if Add AMD GPU support to accelpath (WIP) #205's approach changes, these fixtures/expected outputs may need adjustment.
• This PR was created with the help of AI (an AI coding assistant / GitHub Copilot).

[casparvl]

The main reason for not merging #205 and #214 way back when was that we weren't ready to expose the AMD GPU software yet (i.e. it was only accessible by explicitely setting the EESSI_ACCELERATOR_TARGET_OVERRIDE=accel/amd/gfx90a). We'd like at least the core software to be available, and functional. I guess we should have a discussion on what that core software is - we are up to the level of HIP/RCCL now. Maybe we consider that enough and we expose it? Will discuss in our support meeting.

[casparvl]

Discussed in the support meeting. We think the time is right to actually expose this, i.e. add the archdetect support. Only challenge now is to find the time to review this PR :)

[casparvl]

Just as a note: from the brief look I gave it, this replaces #205, but NOT #214. I.e. if this PR is approved, we should close #205 as 'superseded' by this PR, but we should probably take #214 out of draft mode and still assess it.

Note that #214 in the opening post also states we probably need to update workflows/tests_eessi_extend_module.yml. That is NOT resolved by the current PR, so that gap remains.

[hmeiland]

Just as a note: from the brief look I gave it, this replaces #205, but NOT #214. I.e. if this PR is approved, we should close #205 as 'superseded' by this PR, but we should probably take #214 out of draft mode and still assess it.

Note that #214 in the opening post also states we probably need to update workflows/tests_eessi_extend_module.yml. That is NOT resolved by the current PR, so that gap remains.

on it...

[hmeiland]

Superseded by #273. That PR consolidates AMD GPU accelerator detection in init/eessi_archdetect.sh plus the CI tests (KFD-topology fixtures for MI250X/MI300X/V710 + no-device cases), is rebased on current main, and has all checks passing. Please review #273 instead; closing this one shortly to avoid two competing implementations of the same detection.