Merge branch 'main' into ascend-npu/torchtitan

Author: lowdy1

.github/workflows/_ascend_npu_benchmark.yml

@@ -11,10 +11,6 @@ on:
         required: true
         type: string
         description: "The docker image which will be loaded"
-      device:
-        required: true
-        type: string
-        description: "The device selected to run on"
       torch-artifact:
         required: false
         type: string
@@ -28,13 +24,6 @@ on:
         description: "A token used to create a pull request"
         required: true
 
-# Bash shells do not use ~/.profile or ~/.bashrc so these shells need to be explicitly
-# declared as "shell: bash -el {0}" on steps that need to be properly activated.
-# It's used to activate ascend-toolkit environment variables.
-defaults:
-  run:
-    shell: bash -el {0}
-
 jobs:
   benchmark:
     name: run benchmarks for torch_npu

.github/workflows/_ascend_npu_build_torch.yml

@@ -26,13 +26,6 @@ on:
         description: "The distribution artifact name of torch"
         value: ${{ jobs.build.outputs.dist-name }}
 
-# Bash shells do not use ~/.profile or ~/.bashrc so these shells need to be explicitly
-# declared as "shell: bash -el {0}" on steps that need to be properly activated.
-# It's used to activate ascend-toolkit environment variables.
-defaults:
-  run:
-    shell: bash -el {0}
-
 jobs:
   build:
     name: build torch for ${{ inputs.pr-number && format('#{0}', inputs.pr-number) || inputs.ref }}

.github/workflows/_ascend_npu_build_torch_npu.yml

@@ -11,10 +11,6 @@ on:
         required: true
         type: string
         description: "The docker image which will be used to build"
-      device:
-        required: true
-        type: string
-        description: "The device selected to run on"
       torch-artifact:
         required: false
         type: string
@@ -24,13 +20,6 @@ on:
         description: "The distribution artifact name of torch_npu"
         value: ${{ jobs.build.outputs.dist-name }}
 
-# Bash shells do not use ~/.profile or ~/.bashrc so these shells need to be explicitly
-# declared as "shell: bash -el {0}" on steps that need to be properly activated.
-# It's used to activate ascend-toolkit environment variables.
-defaults:
-  run:
-    shell: bash -el {0}
-
 jobs:
   build:
     name: build torch_npu

.github/workflows/_ascend_npu_ut.yml

@@ -11,10 +11,6 @@ on:
         required: true
         type: string
         description: "The docker image which will be loaded"
-      device:
-        required: true
-        type: string
-        description: "The device selected to run on"
       torch-artifact:
         required: false
         type: string
@@ -24,13 +20,6 @@ on:
         type: string
         description: "The distribution artifact name of torch_npu"
 
-# Bash shells do not use ~/.profile or ~/.bashrc so these shells need to be explicitly
-# declared as "shell: bash -el {0}" on steps that need to be properly activated.
-# It's used to activate ascend-toolkit environment variables.
-defaults:
-  run:
-    shell: bash -el {0}
-
 jobs:
   test:
     name: test torch_npu

.github/workflows/ascend_npu_test.yml

@@ -58,20 +58,6 @@ on:
           - ascendai/cann:latest
         default: "ascendai/cann:latest"
         description: "The docker image which will be loaded"
-      device:
-        required: true
-        type: choice
-        options:
-          - /dev/davinci1
-          - /dev/davinci2
-          - /dev/davinci3
-          - /dev/davinci4
-          - /dev/davinci5
-          - /dev/davinci6
-          - /dev/davinci7
-          - /dev/davinci8
-        default: "/dev/davinci5"
-        description: "The device selected to run on"
 
 # Only cancel the previous runs when triggered by a pull_request event
 #
@@ -100,7 +86,6 @@ jobs:
         id: set-env
         run: |
           echo "runner=${{ github.event.inputs.runner || 'linux-arm64-npu-1' }}" >> $GITHUB_OUTPUT
-          echo "device=${{ github.event.inputs.device || '/dev/davinci5' }}" >> $GITHUB_OUTPUT
           echo "image=${{ github.event.inputs.image || 'ascendai/cann:latest' }}" >> $GITHUB_OUTPUT
 
       # TODO(shink): List ghstack PR's ref
@@ -133,7 +118,6 @@ jobs:
     with:
       runner: ${{ needs.prepare.outputs.runner }}
       image: ${{ needs.prepare.outputs.image }}
-      device: ${{ needs.prepare.outputs.device }}
       torch-artifact: ${{ needs.build-torch.outputs.torch-artifact }}
 
   torchtitan:
@@ -147,9 +131,24 @@ jobs:
       (success() || (needs.build-torch.result == 'skipped' && needs.build.result == 'success'))
     uses: ./.github/workflows/_ascend_npu_torchtitan.yml
     with:
-      runner: "linux-arm64-npu-4"
-      image: "ascendai/cann:8.0.0.beta1-910b-ubuntu22.04-py3.10"
-      device:  ${{ needs.prepare.outputs.device }}
+      runner: ${{ needs.prepare.outputs.runner }}
+      image: ${{ needs.prepare.outputs.image }}
+      torch-artifact: ${{ needs.build-torch.outputs.torch-artifact }}
+      torch-npu-artifact: ${{ needs.build.outputs.torch-npu-artifact }}
+
+  benchmark:
+    name: Run benchmarks
+    needs:
+      - prepare
+      - build-torch
+      - build
+    if: |
+      !cancelled() && github.event_name != 'repository_dispatch' &&
+      (success() || (needs.build-torch.result == 'skipped' && needs.build.result == 'success'))
+    uses: ./.github/workflows/_ascend_npu_benchmark.yml
+    with:
+      runner: ${{ needs.prepare.outputs.runner }}
+      image: ${{ needs.prepare.outputs.image }}
       torch-artifact: ${{ needs.build-torch.outputs.torch-artifact }}
       torch-npu-artifact: ${{ needs.build.outputs.torch-npu-artifact }}
     secrets: