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

.github/workflows/tests_archdetect_amd_gpu.yml

@@ -0,0 +1,80 @@
+# documentation: https://help.github.com/en/articles/workflow-syntax-for-github-actions
+name: Tests for accelerator detection (AMD GPU)
+on:
+  push:
+  pull_request:
+permissions:
+  contents: read # to fetch code (actions/checkout)
+jobs:
+  # Method 1: KFD sysfs topology parsing (no amd-smi / Python required).
+  # Fixtures under tests/archdetect/kfd/<name>/nodes/ are captured from real hardware
+  # and injected via $EESSI_KFD_TOPOLOGY_ROOT, so no AMD GPU is needed in CI.
+  kfd_sysfs:
+    runs-on: ubuntu-24.04
+    strategy:
+      matrix:
+        fixture:
+          - v710      # AMD Radeon Pro V710 (NAVI32), gfx_target_version 110001 -> gfx1101
+          - cpu_only  # only a CPU node (gfx_target_version 0) -> no accelerator, exit 2
+      fail-fast: false
+    steps:
+    - name: checkout
+      uses: actions/checkout@d632683dd7b4114ad314bca15554477dd762a938 # v4.2.0
+
+    - name: test AMD KFD sysfs accelerator detection
+      run: |
+          fixture="${{matrix.fixture}}"
+          root="$PWD/tests/archdetect/kfd/${fixture}/nodes"
+          expected=$(cat "$PWD/tests/archdetect/kfd/${fixture}/expected.output")
+
+          # ensure no real amd-smi leaks in and lets the cpu_only case pass via Method 2
+          export PATH="$(mktemp -d):/usr/bin:/bin"
+
+          export EESSI_KFD_TOPOLOGY_ROOT="$root"
+
+          # first run with debugging enabled, just to show the output
+          ./init/eessi_archdetect.sh -d accelpath || echo "non-zero exit code: $?"
+
+          out=$(./init/eessi_archdetect.sh accelpath || echo "non-zero exit code: $?")
+          if [[ "$out" == "$expected" ]]; then
+              echo "KFD test for '${fixture}' PASSED: '$out'"
+          else
+              echo "KFD test for '${fixture}' FAILED: got '$out', expected '$expected'" >&2
+              exit 1
+          fi
+
+  # Method 2: amd-smi fallback. A fake amd-smi (captured real-hardware output) is placed
+  # on $PATH, and the KFD root is pointed at an empty dir to force the fallback path.
+  amd_smi:
+    runs-on: ubuntu-24.04
+    strategy:
+      matrix:
+        fixture:
+          - v710  # amd-smi reports TARGET_GRAPHICS_VERSION: gfx1101
+      fail-fast: false
+    steps:
+    - name: checkout
+      uses: actions/checkout@d632683dd7b4114ad314bca15554477dd762a938 # v4.2.0
+
+    - name: test AMD amd-smi accelerator detection
+      run: |
+          fixture="${{matrix.fixture}}"
+          expected=$(cat "$PWD/tests/archdetect/amd-smi/${fixture}.output")
+
+          # force Method 1 (KFD) to find nothing so Method 2 (amd-smi) is exercised
+          export EESSI_KFD_TOPOLOGY_ROOT="$(mktemp -d)"
+
+          # put fake amd-smi command in place
+          tmpdir=$(mktemp -d)
+          ln -s "$PWD/tests/archdetect/amd-smi/${fixture}.sh" "$tmpdir/amd-smi"
+          export PATH="$tmpdir:$PATH"
+
+          ./init/eessi_archdetect.sh -d accelpath || echo "non-zero exit code: $?"
+
+          out=$(./init/eessi_archdetect.sh accelpath || echo "non-zero exit code: $?")
+          if [[ "$out" == "$expected" ]]; then
+              echo "amd-smi test for '${fixture}' PASSED: '$out'"
+          else
+              echo "amd-smi test for '${fixture}' FAILED: got '$out', expected '$expected'" >&2
+              exit 1
+          fi

init/eessi_archdetect.sh

@@ -175,11 +175,107 @@ cpupath(){
     fi
 }
 
+nvidia_accelpath() {
+    # Check for NVIDIA GPUs via nvidia-smi command
+    local nvidia_smi
+    nvidia_smi=$(command -v nvidia-smi)
+
+    if [[ $? -eq 0 ]]; then
+        log "DEBUG" "nvidia_accelpath: nvidia-smi command found @ ${nvidia_smi}"
+        local nvidia_smi_out
+        nvidia_smi_out=$(mktemp -p /tmp nvidia_smi_out.XXXXX)
+
+        nvidia-smi --query-gpu=gpu_name,count,driver_version,compute_cap --format=csv,noheader 2>&1 > $nvidia_smi_out
+        if [[ $? -eq 0 ]]; then
+            local nvidia_smi_info=$(head -n 1 $nvidia_smi_out)
+            local cuda_cc=$(echo $nvidia_smi_info | sed 's/, /,/g' | cut -f4 -d, | sed 's/\.//g')
+            log "DEBUG" "nvidia_accelpath: CUDA compute capability '${cuda_cc}' derived from nvidia-smi output '${nvidia_smi_info}'"
+
+            echo "accel/nvidia/cc${cuda_cc}"
+            rm -f $nvidia_smi_out
+            return 0
+        else
+            log "DEBUG" "nvidia_accelpath: nvidia-smi command failed, see output in $nvidia_smi_out"
+            return 3
+        fi
+    else
+        log "DEBUG" "nvidia_accelpath: nvidia-smi command not found"
+        return 2
+    fi
+}
+
+amd_accelpath() {
+    # Method 1: Check for AMD GPUs via KFD sysfs interface (No amd-smi or Python required)
+    # $EESSI_KFD_TOPOLOGY_ROOT overrides the sysfs root so CI can test with captured fixtures.
+    local kfd_nodes="${EESSI_KFD_TOPOLOGY_ROOT:-/sys/devices/virtual/kfd/kfd/topology/nodes}"
+
+    if [[ -d "$kfd_nodes" ]]; then
+        log "DEBUG" "amd_accelpath: KFD sysfs path found @ ${kfd_nodes}"
+        local amdgcn_cc=""
+
+        # ls -1v ensures numeric/version sorting (nodes/0, nodes/1, ..., nodes/10)
+        for node in $(ls -1v "$kfd_nodes" 2>/dev/null); do
+            local prop_file="$kfd_nodes/$node/properties"
+
+            if [[ -f "$prop_file" ]]; then
+                # Extract the integer value. 2>/dev/null suppresses read errors.
+                local gfx_ver=$(grep "^gfx_target_version" "$prop_file" 2>/dev/null | awk '{print $2}')
+
+                # If gfx_ver is non-empty and greater than 0 (0 means it's a CPU node)
+                if [[ -n "$gfx_ver" && "$gfx_ver" -gt 0 ]]; then
+                    local major=$(( (gfx_ver / 10000) % 100 ))
+                    local minor=$(( (gfx_ver / 100) % 100 ))
+                    local step=$(( gfx_ver % 100 ))
+
+                    amdgcn_cc=$(printf "gfx%d%d%x" $major $minor $step)
+                    log "DEBUG" "amd_accelpath: AMDGCN compute capability '${amdgcn_cc}' derived from KFD node ${node}"
+                    break 
+                fi
+            fi
+        done
+
+        if [[ -n "$amdgcn_cc" ]]; then
+            echo "accel/amd/${amdgcn_cc}"
+            return 0
+        fi
+        log "DEBUG" "amd_accelpath: KFD topology found, but no AMD GPUs detected. Falling back to amd-smi."
+    else
+        log "DEBUG" "amd_accelpath: KFD sysfs path not found. Falling back to amd-smi."
+    fi
+
+    # Method 2: Fallback to AMD GPUs via amd-smi command using /tmp files
+    local amd_smi
+    amd_smi=$(command -v amd-smi)
+
+    if [[ $? -eq 0 ]]; then
+        log "DEBUG" "amd_accelpath: amd-smi command found @ ${amd_smi}"
+        local amd_smi_out
+        amd_smi_out=$(mktemp -p /tmp amd_smi_out.XXXXX)
+
+        amd-smi static --asic | grep TARGET_GRAPHICS_VERSION 2>&1 > $amd_smi_out
+        if [[ $? -eq 0 ]]; then
+            local amd_smi_info=$(head -n 1 $amd_smi_out)
+            local amdgcn_cc=$(echo $amd_smi_info | sed 's/.*: //')
+            log "DEBUG" "amd_accelpath: AMDGCN compute capability '${amdgcn_cc}' derived from amd-smi output '${amd_smi_info}'"
+
+            echo "accel/amd/${amdgcn_cc}"
+            rm -f $amd_smi_out
+            return 0
+        else
+            log "DEBUG" "amd_accelpath: amd-smi command failed, see output in $amd_smi_out"
+            return 3
+        fi
+    else
+        log "DEBUG" "amd_accelpath: amd-smi command not found"
+        return 2
+    fi
+}
+
 accelpath() {
     # If EESSI_ACCELERATOR_TARGET_OVERRIDE is set, use it
     log "DEBUG" "accelpath: Override variable set as '$EESSI_ACCELERATOR_TARGET_OVERRIDE' "
     if [ ! -z $EESSI_ACCELERATOR_TARGET_OVERRIDE ]; then
-        # Regex that allows both NVIDIA and AMD overrides
+        # Updated regex to allow both NVIDIA and AMD overrides
         if [[ "$EESSI_ACCELERATOR_TARGET_OVERRIDE" =~ ^accel/(nvidia/cc[0-9]+|amd/gfx[0-9a-f]+)$ ]]; then
             echo "$EESSI_ACCELERATOR_TARGET_OVERRIDE"
             return 0
@@ -189,28 +285,27 @@ accelpath() {
         fi
     fi
 
-    # check for NVIDIA GPUs via nvidia-smi command
-    nvidia_smi=$(command -v nvidia-smi)
+    # 1. Check for NVIDIA GPUs
+    local nv_res
+    nv_res=$(nvidia_accelpath)
     if [[ $? -eq 0 ]]; then
-        log "DEBUG" "accelpath: nvidia-smi command found @ ${nvidia_smi}"
-        nvidia_smi_out=$(mktemp -p /tmp nvidia_smi_out.XXXXX)
-        nvidia-smi --query-gpu=gpu_name,count,driver_version,compute_cap --format=csv,noheader 2>&1 > $nvidia_smi_out
-        if [[ $? -eq 0 ]]; then
-            nvidia_smi_info=$(head -n 1 $nvidia_smi_out)
-            cuda_cc=$(echo $nvidia_smi_info | sed 's/, /,/g' | cut -f4 -d, | sed 's/\.//g')
-            log "DEBUG" "accelpath: CUDA compute capability '${cuda_cc}' derived from nvidia-smi output '${nvidia_smi_info}'"
-            res="accel/nvidia/cc${cuda_cc}"
-            log "DEBUG" "accelpath: result: ${res}"
-            echo $res
-            rm -f $nvidia_smi_out
-        else
-            log "DEBUG" "accelpath: nvidia-smi command failed, see output in $nvidia_smi_out"
-            exit 3
-        fi
-    else
-        log "DEBUG" "accelpath: nvidia-smi command not found"
-        exit 2
+        log "DEBUG" "accelpath: result: ${nv_res}"
+        echo "$nv_res"
+        return 0
     fi
+
+    # 2. Check for AMD GPUs
+    local amd_res
+    amd_res=$(amd_accelpath)
+    if [[ $? -eq 0 ]]; then
+        log "DEBUG" "accelpath: result: ${amd_res}"
+        echo "$amd_res"
+        return 0
+    fi
+
+    # 3. Fail gracefully if neither is found
+    log "DEBUG" "accelpath: No supported accelerators found on this system."
+    exit 2
 }
 
 # Parse command line arguments

tests/archdetect/amd-smi/none.output

@@ -0,0 +1 @@
+non-zero exit code: 2

tests/archdetect/amd-smi/v710.output

@@ -0,0 +1 @@
+accel/amd/gfx1101

tests/archdetect/amd-smi/v710.sh

@@ -0,0 +1,19 @@
+#!/bin/bash
+# output from AMD Radeon Pro V710 (NAVI32, gfx1101) system,
+# produced by: amd-smi static --asic
+cat <<'ASIC'
+GPU: 0
+    ASIC:
+        MARKET_NAME: NAVI32
+        VENDOR_ID: 0x1002
+        VENDOR_NAME: Advanced Micro Devices Inc. [AMD/ATI]
+        SUBVENDOR_ID: 0x1002
+        DEVICE_ID: 0x7461
+        SUBSYSTEM_ID: 0x0e34
+        REV_ID: 0x00
+        ASIC_SERIAL: 0xF7BD1622840AAC82
+        OAM_ID: N/A
+        NUM_COMPUTE_UNITS: 54
+        TARGET_GRAPHICS_VERSION: gfx1101
+ASIC
+exit 0

tests/archdetect/kfd/cpu_only/expected.output

@@ -0,0 +1 @@
+non-zero exit code: 2

tests/archdetect/kfd/cpu_only/nodes/0/properties

@@ -0,0 +1,30 @@
+cpu_cores_count 24
+simd_count 0
+mem_banks_count 1
+caches_count 0
+io_links_count 1
+p2p_links_count 0
+cpu_core_id_base 0
+simd_id_base 0
+max_waves_per_simd 0
+lds_size_in_kb 0
+gds_size_in_kb 0
+num_gws 0
+wave_front_size 0
+array_count 0
+simd_arrays_per_engine 0
+cu_per_simd_array 0
+simd_per_cu 0
+max_slots_scratch_cu 0
+gfx_target_version 0
+vendor_id 0
+device_id 0
+location_id 0
+domain 0
+drm_render_minor 0
+hive_id 0
+num_sdma_engines 0
+num_sdma_xgmi_engines 0
+num_sdma_queues_per_engine 0
+num_cp_queues 0
+max_engine_clk_ccompute 0

tests/archdetect/kfd/v710/expected.output

@@ -0,0 +1 @@
+accel/amd/gfx1101

tests/archdetect/kfd/v710/nodes/0/properties

@@ -0,0 +1,30 @@
+cpu_cores_count 24
+simd_count 0
+mem_banks_count 1
+caches_count 0
+io_links_count 1
+p2p_links_count 0
+cpu_core_id_base 0
+simd_id_base 0
+max_waves_per_simd 0
+lds_size_in_kb 0
+gds_size_in_kb 0
+num_gws 0
+wave_front_size 0
+array_count 0
+simd_arrays_per_engine 0
+cu_per_simd_array 0
+simd_per_cu 0
+max_slots_scratch_cu 0
+gfx_target_version 0
+vendor_id 0
+device_id 0
+location_id 0
+domain 0
+drm_render_minor 0
+hive_id 0
+num_sdma_engines 0
+num_sdma_xgmi_engines 0
+num_sdma_queues_per_engine 0
+num_cp_queues 0
+max_engine_clk_ccompute 0

tests/archdetect/kfd/v710/nodes/1/properties

@@ -0,0 +1,38 @@
+cpu_cores_count 0
+simd_count 108
+mem_banks_count 1
+caches_count 116
+io_links_count 1
+p2p_links_count 0
+cpu_core_id_base 0
+simd_id_base 2147487744
+max_waves_per_simd 16
+lds_size_in_kb 64
+gds_size_in_kb 0
+num_gws 64
+wave_front_size 32
+array_count 6
+simd_arrays_per_engine 2
+cu_per_simd_array 10
+simd_per_cu 2
+max_slots_scratch_cu 32
+gfx_target_version 110001
+vendor_id 4098
+device_id 29793
+location_id 0
+domain 2
+drm_render_minor 128
+hive_id 0
+num_sdma_engines 2
+num_sdma_xgmi_engines 0
+num_sdma_queues_per_engine 6
+num_cp_queues 14
+max_engine_clk_fcompute 2585
+local_mem_size 0
+fw_version 2460
+capability 734503552
+debug_prop 1495
+sdma_fw_version 25
+unique_id 17851448835466505346
+num_xcc 1
+max_engine_clk_ccompute 0