try gau's commit

huggingface · Nov 18, 2023 · 050cbbd · 050cbbd
1 parent 0ef8c19
commit 050cbbd
Showing 1 changed file with 20 additions and 22 deletions.
diff --git a/.github/workflows/self-scheduled-amd.yml b/.github/workflows/self-scheduled-amd.yml
@@ -43,18 +43,19 @@ jobs:
         machine_type: [single-gpu, multi-gpu]
     runs-on: [self-hosted, docker-gpu, amd-gpu, '${{ matrix.machine_type }}', '${{ inputs.gpu_flavor }}']
     container:
-      image: huggingface/transformers-pytorch-amd-gpu
+      image: huggingface/transformers-pytorch-amd-gpu-test
       options: --device /dev/kfd --device /dev/dri --env ROCR_VISIBLE_DEVICES --shm-size "16gb" --ipc host -v /mnt/cache/.cache/huggingface:/mnt/cache/
     steps:
       - name: ROCM-SMI
         run: |
           rocm-smi
-      - name: ROCM-INFO
-        run: |
-          rocminfo  | grep "Agent" -A 14
       - name: Show ROCR environment
         run: |
           echo "ROCR: $ROCR_VISIBLE_DEVICES"
+          echo "ROCR: $ROCR_VISIBLE_DEVICES"
+      - name: ROCMINFO
+        run: |
+          rocminfo  | grep "Agent" -A 14
 
   setup:
     name: Setup
@@ -64,7 +65,7 @@ jobs:
         machine_type: [single-gpu, multi-gpu]
     runs-on: [self-hosted, docker-gpu, amd-gpu, '${{ matrix.machine_type }}', '${{ inputs.gpu_flavor }}']
     container:
-      image: huggingface/transformers-pytorch-amd-gpu
+      image: huggingface/transformers-pytorch-amd-gpu-test
       options: --device /dev/kfd --device /dev/dri --env ROCR_VISIBLE_DEVICES --shm-size "16gb" --ipc host -v /mnt/cache/.cache/huggingface:/mnt/cache/
     outputs:
       matrix: ${{ steps.set-matrix.outputs.matrix }}
@@ -94,12 +95,12 @@ jobs:
       - name: ROCM-SMI
         run: |
           rocm-smi
-      - name: ROCM-INFO
-        run: |
-          rocminfo  | grep "Agent" -A 14
       - name: Show ROCR environment
         run: |
           echo "ROCR: $ROCR_VISIBLE_DEVICES"
+      - name: ROCMINFO
+        run: |
+          rocminfo  | grep "Agent" -A 14
 
   run_tests_single_gpu:
     name: Single GPU tests
@@ -111,7 +112,7 @@ jobs:
         machine_type: [single-gpu]
     runs-on: [self-hosted, docker-gpu, amd-gpu, '${{ matrix.machine_type }}', '${{ inputs.gpu_flavor }}']
     container:
-      image: huggingface/transformers-pytorch-amd-gpu
+      image: huggingface/transformers-pytorch-amd-gpu-test
       options: --device /dev/kfd --device /dev/dri --env ROCR_VISIBLE_DEVICES --shm-size "16gb" --ipc host -v /mnt/cache/.cache/huggingface:/mnt/cache/
     needs: setup
     steps:
@@ -126,6 +127,10 @@ jobs:
           echo "$matrix_folders"
           echo "matrix_folders=$matrix_folders" >> $GITHUB_ENV
 
+      - name: ROCMINFO
+        run: |
+          rocminfo  | grep "Agent" -A 14
+
       - name: Update clone
         working-directory: /transformers
         run: git fetch && git checkout ${{ github.sha }}
@@ -137,9 +142,6 @@ jobs:
       - name: ROCM-SMI
         run: |
           rocm-smi
-      - name: ROCM-INFO
-        run: |
-          rocminfo  | grep "Agent" -A 14
       - name: Show ROCR environment
         run: |
           echo "ROCR: $ROCR_VISIBLE_DEVICES"
@@ -179,7 +181,7 @@ jobs:
         machine_type: [multi-gpu]
     runs-on: [self-hosted, docker-gpu, amd-gpu, '${{ matrix.machine_type }}', '${{ inputs.gpu_flavor }}']
     container:
-      image: huggingface/transformers-pytorch-amd-gpu
+      image: huggingface/transformers-pytorch-amd-gpu-test
       options: --device /dev/kfd --device /dev/dri --env ROCR_VISIBLE_DEVICES --shm-size "16gb" --ipc host -v /mnt/cache/.cache/huggingface:/mnt/cache/
     needs: setup
     steps:
@@ -205,9 +207,11 @@ jobs:
       - name: ROCM-SMI
         run: |
           rocm-smi
-      - name: ROCM-INFO
+
+      - name: ROCMINFO
         run: |
           rocminfo  | grep "Agent" -A 14
+
       - name: Show ROCR environment
         run: |
           echo "ROCR: $ROCR_VISIBLE_DEVICES"
@@ -245,7 +249,7 @@ jobs:
         machine_type: [single-gpu]
     runs-on: [self-hosted, docker-gpu, amd-gpu, '${{ matrix.machine_type }}', '${{ inputs.gpu_flavor }}']
     container:
-      image: huggingface/transformers-pytorch-amd-gpu
+      image: huggingface/transformers-pytorch-amd-gpu-test
       options: --device /dev/kfd --device /dev/dri --env ROCR_VISIBLE_DEVICES --shm-size "16gb" --ipc host -v /mnt/cache/.cache/huggingface:/mnt/cache/
     needs: setup
     steps:
@@ -260,9 +264,6 @@ jobs:
       - name: ROCM-SMI
         run: |
           rocm-smi
-      - name: ROCM-INFO
-        run: |
-          rocminfo  | grep "Agent" -A 14
       - name: Show ROCR environment
         run: |
           echo "ROCR: $ROCR_VISIBLE_DEVICES"
@@ -302,7 +303,7 @@ jobs:
         machine_type: [single-gpu, multi-gpu]
     runs-on: [self-hosted, docker-gpu, amd-gpu, '${{ matrix.machine_type }}', '${{ inputs.gpu_flavor }}']
     container:
-      image: huggingface/transformers-pytorch-amd-gpu
+      image: huggingface/transformers-pytorch-amd-gpu-test
       options: --device /dev/kfd --device /dev/dri --env ROCR_VISIBLE_DEVICES --shm-size "16gb" --ipc host -v /mnt/cache/.cache/huggingface:/mnt/cache/
     needs: setup
     steps:
@@ -317,9 +318,6 @@ jobs:
       - name: ROCM-SMI
         run: |
           rocm-smi
-      - name: ROCM-INFO
-        run: |
-          rocminfo  | grep "Agent" -A 14
       - name: Show ROCR environment
         run: |
           echo "ROCR: $ROCR_VISIBLE_DEVICES"