Convert ci-shark-ai.yml to use pkgci_shark_ai.yml so that we only…

… build packages once (#780)

This builds on #625, #589 to make progress on issue #584.

This adds a pkgci.yml to run multiple package-based CI tasks after
building package using Scott's changes in #667. This gives us the
following benefits:

* Integration test workflows are faster because they now use dev
packages, without needing to build them from source or use editable
installs. Also, if more integration tests are added, they can reuse the
built packages.
* Users and developers can access the same dev packages to reproduce CI
results
* Only one runner needs the build requirements (potentially including
clang, ninja, CMake, Rust, etc.), other runners only need Python.

This also switches to using uv to create venvs, which is faster.

This PR brings shortfin CPU LLM CI time to roughly half an hour on the
mi250 runner to a few seconds of package build (fast due to caching) and
around 5 minutes of testing.

---------

Co-authored-by: Scott Todd <[email protected]>

.github/workflows/pkgci.yml

@@ -0,0 +1,39 @@
+# Copyright 2024 Advanced Micro Devices, Inc.
+#
+# Licensed under the Apache License v2.0 with LLVM Exceptions.
+# See https://llvm.org/LICENSE.txt for license information.
+# SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+
+name: PkgCI
+
+on:
+  workflow_dispatch:
+  pull_request:
+  push:
+    branches:
+      - main
+
+permissions:
+  contents: read
+
+concurrency:
+  # A PR number if a pull request and otherwise the commit hash. This cancels
+  # queued and in-progress runs for the same PR (presubmit) or commit
+  # (postsubmit). The workflow name is prepended to avoid conflicts between
+  # different workflows.
+  group: ${{ github.workflow }}-${{ github.event.number || github.sha }}
+  cancel-in-progress: true
+
+jobs:
+  build_packages:
+    name: Build Packages
+    uses: ./.github/workflows/build_packages.yml
+    permissions:
+      contents: write
+    with:
+      build_type: "dev"
+
+  test_shark_ai:
+    name: Test shark-ai
+    needs: [build_packages]
+    uses: ./.github/workflows/pkgci_shark_ai.yml

.github/workflows/pkgci_shark_ai.yml

@@ -0,0 +1,95 @@
+# Copyright 2024 Advanced Micro Devices, Inc.
+#
+# Licensed under the Apache License v2.0 with LLVM Exceptions.
+# See https://llvm.org/LICENSE.txt for license information.
+# SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+
+name: PkgCI - shark-ai
+
+on:
+  workflow_call:
+    inputs:
+      artifact_run_id:
+        type: string
+        default: ""
+  workflow_dispatch:
+    inputs:
+      artifact_run_id:
+        type: string
+        description: "Id for a workflow run that produced dev packages"
+        default: ""
+
+jobs:
+  test_shortfin_llm_server:
+    name: "Integration Tests - Shortfin LLM Server"
+    strategy:
+      matrix:
+        version: [3.11]
+      fail-fast: false
+    runs-on: mi300x-4
+    # runs-on: ubuntu-latest # everything else works but this throws an "out of resources" during model loading
+    # TODO: make a copy of this that runs on standard runners with tiny llama instead of a 8b model
+    defaults:
+      run:
+        shell: bash
+    env:
+      PACKAGE_DOWNLOAD_DIR: ${{ github.workspace }}/.packages
+      VENV_DIR: ${{ github.workspace }}/.venv
+    steps:
+      - name: "Checkout Code"
+        uses: actions/checkout@11bd71901bbe5b1630ceea73d27597364c9af683 # v4.2.2
+      - name: "Setting up Python"
+        id: setup_python
+        uses: actions/setup-python@0b93645e9fea7318ecaed2b359559ac225c90a2b # v5.3.0
+        with:
+          python-version: ${{matrix.version}}
+
+      - name: Set Python version without dot
+        run: |
+          echo "PY_VERSION_NO_DOT=$(echo ${{ matrix.version }} | tr -d '.')" >> $GITHUB_ENV
+
+      - name: Setup UV caching
+        run: |
+          CACHE_DIR="${GITHUB_WORKSPACE}/.uv-cache"
+          echo "UV_CACHE_DIR=${CACHE_DIR}" >> $GITHUB_ENV
+          mkdir -p "${CACHE_DIR}"
+
+      - name: Cache UV packages
+        uses: actions/cache@6849a6489940f00c2f30c0fb92c6274307ccb58a # v4.1.2
+        with:
+          path: .uv-cache
+          key: ${{ runner.os }}-uv-py${{ matrix.version }}-${{ hashFiles('requirements-iree-pinned.txt', 'pytorch-cpu-requirements.txt', 'sharktank/requirements.txt', 'sharktank/requirements-tests.txt', 'shortfin/requirements-tests.txt') }}
+
+      - name: Download sharktank artifacts
+        uses: actions/download-artifact@fa0a91b85d4f404e444e00e005971372dc801d16 # v4.1.8
+        with:
+          name: snapshot-sharktank-linux-x86_64-cp${{ env.PY_VERSION_NO_DOT }}-cp${{ env.PY_VERSION_NO_DOT }}
+          path: ${{ env.PACKAGE_DOWNLOAD_DIR }}
+
+      - name: Download shortfin artifacts
+        uses: actions/download-artifact@fa0a91b85d4f404e444e00e005971372dc801d16 # v4.1.8
+        with:
+          name: snapshot-shortfin-linux-x86_64-cp${{ env.PY_VERSION_NO_DOT }}-cp${{ env.PY_VERSION_NO_DOT }}
+          path: ${{ env.PACKAGE_DOWNLOAD_DIR }}
+
+      - name: Download shark-ai artifacts
+        uses: actions/download-artifact@fa0a91b85d4f404e444e00e005971372dc801d16 # v4.1.8
+        with:
+          name: snapshot-shark-ai-linux-x86_64-cp${{ env.PY_VERSION_NO_DOT }}-cp${{ env.PY_VERSION_NO_DOT }}
+          path: ${{ env.PACKAGE_DOWNLOAD_DIR }}
+
+      - name: Setup venv
+        run: |
+          ./build_tools/pkgci/setup_venv.py ${VENV_DIR} \
+            --artifact-path=${PACKAGE_DOWNLOAD_DIR} \
+            --fetch-gh-workflow=${{ inputs.artifact_run_id }}
+
+      - name: Install pinned IREE packages
+        run: |
+          source ${VENV_DIR}/bin/activate
+          uv pip install -r requirements-iree-pinned.txt
+
+      - name: Run LLM Integration Tests
+        run: |
+          source ${VENV_DIR}/bin/activate
+          pytest -v -s app_tests/integration_tests/llm/shortfin --log-cli-level=INFO