upload NIGHTLY assets to GCP

.github/actions/nm-cp-assets/action.yml

@@ -0,0 +1,23 @@
+name: cp assets
+description: "cp whl and tarfile to Google storage 'neuralmagic-public-pypi/dist'"
+inputs:
+  python:
+    description: 'python version, e.g. 3.10.12'
+    required: true
+runs:
+  using: composite
+  steps:
+    - id: mv_assets
+      run: |
+        VERSION_BASE=$(./.github/scripts/convert-version ${{ inputs.python }})
+        WHL=$(find assets -type f -name "*nm_vllm*${VERSION_BASE}*.whl")
+        WHL_FILE=$(basename ${WHL})
+        echo "whl: ${WHL}"
+        echo "whl_file: ${WHL_FILE}"
+        TAR=$(find assets -path "*${{ inputs.python }}-nm-vllm*.tar.gz" -type f -name "nm-vllm*.tar.gz")
+        TAR_FILE=$(basename ${TAR})
+        echo "tar: ${TAR}"
+        echo "tar_file: ${TAR_FILE}"
+        gcloud storage cp ${WHL} gs://neuralmagic-public-pypi/dist/${WHL_FILE}
+        gcloud storage cp ${TAR} gs://neuralmagic-public-pypi/dist/${TAR_FILE}
+      shell: bash

.github/workflows/build-test.yml

@@ -94,29 +94,29 @@ jobs:
             test_skip_list: ${{ inputs.test_skip_list }}
         secrets: inherit
 
-    TEST-MULTI:
-        needs: [BUILD]
-        if: success() && contains(fromJSON('["NIGHTLY", "WEEKLY", "RELEASE"]'), inputs.wf_category)
-        uses: ./.github/workflows/test.yml
-        with:
-            test_label: ${{ inputs.test_label_multi }}
-            timeout: ${{ inputs.test_timeout }}
-            gitref: ${{ github.ref }}
-            python: ${{ inputs.python }}
-            whl: ${{ needs.BUILD.outputs.whl }}
-            test_skip_list: ${{ inputs.test_skip_list }}
-        secrets: inherit
+    # TODO: re-enable
+    # TEST-MULTI:
+    #     needs: [BUILD]
+    #     if: success() && contains(fromJSON('["NIGHTLY", "WEEKLY", "RELEASE"]'), inputs.wf_category)
+    #     uses: ./.github/workflows/test.yml
+    #     with:
+    #         test_label: ${{ inputs.test_label_multi }}
+    #         timeout: ${{ inputs.test_timeout }}
+    #         gitref: ${{ github.ref }}
+    #         python: ${{ inputs.python }}
+    #         whl: ${{ needs.BUILD.outputs.whl }}
+    #         test_skip_list: ${{ inputs.test_skip_list }}
+    #     secrets: inherit
 
-    PUBLISH:
-        needs: [TEST-SOLO, TEST-MULTI]
-        uses: ./.github/workflows/nm-publish.yml
+    UPLOAD:
+        needs: [TEST-SOLO]
+        if: contains(fromJSON('["NIGHTLY", "WEEKLY", "RELEASE"]'), inputs.wf_category)
+        uses: ./.github/workflows/nm-upload-assets-to-gcp.yml
         with:
             label: ${{ inputs.build_label }}
             timeout: ${{ inputs.build_timeout }}
             gitref: ${{ github.ref }}
             python: ${{ inputs.python }}
-            whl: ${{ needs.BUILD.outputs.whl }}
-            tarfile: ${{ needs.BUILD.outputs.tarfile }}
         secrets: inherit
 
     BENCHMARK:

.github/workflows/nightly.yml

@@ -27,10 +27,10 @@ jobs:
             test_label_solo: aws-test-a10g-24G
             test_label_multi: ignore
             test_timeout: 480
-            test_skip_list: neuralmagic/tests/skip-for-nightly.txt
+            test_skip_list: neuralmagic/tests/skip-for-remote-push-tmp.txt
 
             benchmark_label: aws-test-a10g-24G
-            benchmark_config_list_file: ./.github/data/nm_benchmark_nightly_configs_list.txt
+            benchmark_config_list_file: ./.github/data/nm_benchmark_remote_push_configs_list.txt
             benchmark_timeout: 720
             push_benchmark_results_to_gh_pages: "${{ github.event_name == 'schedule' || inputs.push_benchmark_results_to_gh_pages }}"
         secrets: inherit
@@ -45,10 +45,10 @@ jobs:
             test_label_solo: aws-test-a10g-24G
             test_label_multi: ignore
             test_timeout: 480
-            test_skip_list: neuralmagic/tests/skip-for-nightly.txt
+            test_skip_list: neuralmagic/tests/skip-for-remote-push-tmp.txt
 
             benchmark_label: aws-test-a10g-24G
-            benchmark_config_list_file: ./.github/data/nm_benchmark_nightly_configs_list.txt
+            benchmark_config_list_file: ./.github/data/nm_benchmark_remote_push_configs_list.txt
             benchmark_timeout: 720
             push_benchmark_results_to_gh_pages: "${{ github.event_name == 'schedule' || inputs.push_benchmark_results_to_gh_pages }}"
         secrets: inherit
@@ -60,13 +60,13 @@ jobs:
             python: 3.10.12
             gitref: ${{ github.ref }}
 
-            test_label_solo: aws-test-a10g-24G
-            test_label_multi: aws-test-4-a10g-96G
+            test_label_solo: aws-avx2-32G-a10g-24G
+            test_label_multi: ignore
             test_timeout: 480
-            test_skip_list: neuralmagic/tests/skip-for-nightly.txt
+            test_skip_list: neuralmagic/tests/skip-for-remote-push-tmp.txt
 
             benchmark_label: aws-test-a10g-24G
-            benchmark_config_list_file: ./.github/data/nm_benchmark_nightly_configs_list.txt
+            benchmark_config_list_file: ./.github/data/nm_benchmark_remote_push_configs_list.txt
             benchmark_timeout: 720
             push_benchmark_results_to_gh_pages: "${{ github.event_name == 'schedule' || inputs.push_benchmark_results_to_gh_pages }}"
         secrets: inherit
@@ -81,10 +81,10 @@ jobs:
             test_label_solo: aws-test-a10g-24G
             test_label_multi: ignore
             test_timeout: 480
-            test_skip_list: neuralmagic/tests/skip-for-nightly.txt
+            test_skip_list: neuralmagic/tests/skip-for-remote-push-tmp.txt
 
             benchmark_label: aws-test-a10g-24G
-            benchmark_config_list_file: ./.github/data/nm_benchmark_nightly_configs_list.txt
+            benchmark_config_list_file: ./.github/data/nm_benchmark_remote_push_configs_list.txt
             benchmark_timeout: 720
             push_benchmark_results_to_gh_pages: "${{ github.event_name == 'schedule' || inputs.push_benchmark_results_to_gh_pages }}"
         secrets: inherit

.github/workflows/nm-upload-assets-to-gcp.yml

@@ -0,0 +1,72 @@
+name: nm upload assets to GCP
+run-name: ${{ github.actor }} publishing whl on branch '${{ github.ref}}'
+on:
+
+    workflow_call:
+        inputs:
+            label:
+                description: "requested runner label (specifies instance)"
+                type: string
+                required: true
+            timeout:
+                description: "time limit for run in minutes "
+                type: string
+                required: true
+            gitref:
+                description: 'git commit hash or branch name'
+                type: string
+                required: true
+            python:
+                description: "python version, e.g. 3.10.12"
+                type: string
+                required: true
+
+jobs:
+
+    PUBLISH:
+        runs-on: ${{ inputs.label }}
+        timeout-minutes: ${{ fromJson(inputs.timeout) }}
+
+        permissions:
+            contents: 'read'
+            id-token: 'write'
+
+        steps:
+
+            - name: checkout
+              id: checkout
+              uses: actions/checkout@v4
+              with:
+                fetch-depth: 0
+                ref: ${{ inputs.gitref }}
+                submodules: recursive
+
+            - name: setenv
+              id: setenv
+              uses: ./.github/actions/nm-set-env/
+
+            - name: download assets
+              id: download_whl
+              uses: actions/download-artifact@v4
+              with:
+                  path: assets
+
+            # GCP
+            - name: 'Authenticate to Google Cloud'
+              id: auth
+              uses: google-github-actions/[email protected]
+              with:
+                  project_id: ${{ secrets.GCP_PROJECT }}
+                  workload_identity_provider: ${{ secrets.GCP_WORKLOAD_IDENTITY_PROVIDER }}
+                  service_account: ${{ secrets.GCP_SA }}
+
+            - name: 'Set up Cloud SDK'
+              uses: 'google-github-actions/setup-gcloud@v2'
+              with:
+                  version: '>= 473.0.0'
+
+            - name: cp assets
+              id: cp-assets
+              uses: ./.github/actions/nm-cp-assets/
+              with:
+                  python: ${{ inputs.python }}

.github/workflows/release.yml

@@ -21,7 +21,7 @@ jobs:
       gitref: ${{ github.ref }}
 
       test_label_solo: aws-test-a10g-24G
-      test_label_multi: aws-test-4-a10g-96G
+      test_label_multi: ignore
       test_timeout: 720
       test_skip_list: neuralmagic/tests/skip-for-release.txt
 
@@ -39,7 +39,7 @@ jobs:
       gitref: ${{ github.ref }}
 
       test_label_solo: aws-test-a10g-24G
-      test_label_multi: aws-test-4-a10g-96G
+      test_label_multi: ignore
       test_timeout: 720
       test_skip_list: neuralmagic/tests/skip-for-release.txt
 
@@ -57,7 +57,7 @@ jobs:
       gitref: ${{ github.ref }}
 
       test_label_solo: aws-test-a10g-24G
-      test_label_multi: aws-test-4-a10g-96G
+      test_label_multi: ignore
       test_timeout: 720
       test_skip_list: neuralmagic/tests/skip-for-release.txt
 
@@ -75,7 +75,7 @@ jobs:
       gitref: ${{ github.ref }}
 
       test_label_solo: aws-test-a10g-24G
-      test_label_multi: aws-test-4-a10g-96G
+      test_label_multi: ignore
       test_timeout: 720
       test_skip_list: neuralmagic/tests/skip-for-release.txt