Use ort-nightly build for genai gha ci

.github/workflows/linux-cpu-x64-build.yml

@@ -4,10 +4,10 @@ concurrency:
   group: ${{ github.workflow }}-${{ github.head_ref || github.run_id }}
   cancel-in-progress: true
 env:
-  ort_dir: "onnxruntime-linux-x64-1.17.3"
-  ort_zip: "onnxruntime-linux-x64-1.17.3.tgz"
-  ort_url: "https://github.com/microsoft/onnxruntime/releases/download/v1.17.3/onnxruntime-linux-x64-1.17.3.tgz"
-
+  ORT_NIGHTLY_REST_API: "https://feeds.dev.azure.com/aiinfra/PublicPackages/_apis/packaging/Feeds/ORT-Nightly/packages?packageNameQuery=Microsoft.ML.OnnxRuntime&api-version=6.0-preview.1"
+  ORT_PACKAGE_NAME: "Microsoft.ML.OnnxRuntime"
+  ORT_NIGHTLY_SOURCE: "https://aiinfra.pkgs.visualstudio.com/PublicPackages/_packaging/ORT-Nightly/nuget/v3/index.json"
+  NUGET_EXE: "mono /usr/local/bin/nuget.exe"
 jobs:
   linux_cpu_x64:
     runs-on: [ "self-hosted", "1ES.Pool=onnxruntime-genai-Ubuntu2204-AMD-CPU" ]
@@ -16,19 +16,49 @@ jobs:
         uses: actions/checkout@v4
         with:
           submodules: true
+      - name: install Mono and Nuget
+        run: |
+          sudo apt install ca-certificates gnupg
+          sudo gpg --homedir /tmp --no-default-keyring --keyring /usr/share/keyrings/mono-official-archive-keyring.gpg --keyserver hkp://keyserver.ubuntu.com:80 --recv-keys 3FA7E0328081BFF6A14DA29AA6A19B38D3D831EF
+          echo "deb [signed-by=/usr/share/keyrings/mono-official-archive-keyring.gpg] https://download.mono-project.com/repo/ubuntu stable-focal main" | sudo tee /etc/apt/sources.list.d/mono-official-stable.list
+          sudo apt update
+          sudo apt install -y mono-devel
+          sudo curl -o /usr/local/bin/nuget.exe https://dist.nuget.org/win-x86-commandline/latest/nuget.exe
+          sudo chmod +x /usr/local/bin/nuget.exe
+
+      - name: Install jq and dotnet
+        run: |
+          wget https://packages.microsoft.com/config/ubuntu/22.04/packages-microsoft-prod.deb -O packages-microsoft-prod.deb
+          sudo dpkg -i packages-microsoft-prod.deb
+          rm packages-microsoft-prod.deb
+          sudo apt-get update && sudo apt-get install -y dotnet-sdk-8.0 jq
 
-      - name: Download OnnxRuntime
+      - name: Get the Latest OnnxRuntime Nightly Version
         run: |
-          curl -L -o ${{ env.ort_zip }} ${{ env.ort_url }} 
+          ORT_NIGHTLY_VERSION=$(curl -s "${{ env.ORT_NIGHTLY_REST_API }}" | jq -r '.value[0].versions[0].normalizedVersion')
+          echo "$ORT_NIGHTLY_VERSION" 
+          echo "ORT_NIGHTLY_VERSION=$ORT_NIGHTLY_VERSION" >> $GITHUB_ENV
 
-      - name: Unzip OnnxRuntime
+      - name: Download OnnxRuntime Nightly
         run: |
-          tar -xzf ${{ env.ort_zip }}
-          rm ${{ env.ort_zip }}
+          ${{ env.NUGET_EXE }} install ${{ env.ORT_PACKAGE_NAME }} -version ${{ env.ORT_NIGHTLY_VERSION }} -x
+        continue-on-error: true
+
+      - name: list files
+        shell: bash
+        run: |
+          ls -l
+          ls -R ${{ env.ORT_PACKAGE_NAME }}
+        continue-on-error: true
 
-      - name: Rename OnnxRuntime to ort
+# TODO: Find out why do we need to to have libonnxruntime.so.$ort_version
+      - name: Extra OnnxRuntime library and header files
         run: |
-          mv ${{ env.ort_dir }} ort
+          mkdir -p ort/lib
+          mv ${{ env.ORT_PACKAGE_NAME }}/build/native/include ort/
+          mv ${{ env.ORT_PACKAGE_NAME }}/runtimes/linux-x64/native/* ort/lib/
+          ort_version=$(echo ${{ env.ORT_NIGHTLY_VERSION }} | cut -d- -f1-1)
+          cp ort/lib/libonnxruntime.so ort/lib/libonnxruntime.so.$ort_version
 
       - name: Build with CMake and GCC
         run: |

.github/workflows/linux-gpu-x64-build.yml

@@ -6,9 +6,11 @@ concurrency:
   cancel-in-progress: true
 
 env:
-  ort_dir: "onnxruntime-linux-x64-gpu-1.17.3"
-  ort_zip: "onnxruntime-linux-x64-gpu-1.17.3.tgz"
-  ort_url: "https://github.com/microsoft/onnxruntime/releases/download/v1.17.3/onnxruntime-linux-x64-gpu-1.17.3.tgz"
+  ORT_NIGHTLY_REST_API: "https://feeds.dev.azure.com/aiinfra/PublicPackages/_apis/packaging/Feeds/ORT-Nightly/packages?packageNameQuery=Microsoft.ML.OnnxRuntime.Gpu.Linux&api-version=6.0-preview.1"
+  ORT_PACKAGE_NAME: Microsoft.ML.OnnxRuntime.Gpu.Linux
+  ORT_NIGHTLY_SOURCE: "https://aiinfra.pkgs.visualstudio.com/PublicPackages/_packaging/ORT-Nightly/nuget/v3/index.json"
+  NUGET_EXE: "mono /usr/local/bin/nuget.exe"
+
 
 jobs:
   linux-cuda-x64-build:
@@ -29,19 +31,49 @@ jobs:
           clean: true
           path: manylinux
           submodules: true
+      - name: install Mono and Nuget
+        run: |
+          sudo apt install ca-certificates gnupg
+          sudo gpg --homedir /tmp --no-default-keyring --keyring /usr/share/keyrings/mono-official-archive-keyring.gpg --keyserver hkp://keyserver.ubuntu.com:80 --recv-keys 3FA7E0328081BFF6A14DA29AA6A19B38D3D831EF
+          echo "deb [signed-by=/usr/share/keyrings/mono-official-archive-keyring.gpg] https://download.mono-project.com/repo/ubuntu stable-focal main" | sudo tee /etc/apt/sources.list.d/mono-official-stable.list
+          sudo apt update
+          sudo apt install -y mono-devel
+          sudo curl -o /usr/local/bin/nuget.exe https://dist.nuget.org/win-x86-commandline/latest/nuget.exe
+          sudo chmod +x /usr/local/bin/nuget.exe
+
+      - name: Install jq and dotnet
+        run: |
+          wget https://packages.microsoft.com/config/ubuntu/20.04/packages-microsoft-prod.deb -O packages-microsoft-prod.deb
+          sudo dpkg -i packages-microsoft-prod.deb
+          rm packages-microsoft-prod.deb
+          sudo apt-get update && sudo apt-get install -y dotnet-sdk-8.0 jq
 
       - name: Download OnnxRuntime
         run: |
-          curl -L -o ${{ env.ort_zip }} ${{ env.ort_url }}
+          ORT_NIGHTLY_VERSION=$(curl -s "${{ env.ORT_NIGHTLY_REST_API }}" | jq -r '.value[0].versions[0].normalizedVersion')
+          echo "$ORT_NIGHTLY_VERSION" 
+          echo "ORT_NIGHTLY_VERSION=$ORT_NIGHTLY_VERSION" >> $GITHUB_ENV
 
-      - name: Unzip OnnxRuntime
+      - name: Download OnnxRuntime Nightly
+        run: |
+          mono /usr/local/bin/nuget.exe install ${{ env.ORT_PACKAGE_NAME }} -version ${{ env.ORT_NIGHTLY_VERSION }} -x
+        continue-on-error: true
+      - name: list files
+        shell: bash
         run: |
-          tar -xzf ${{ env.ort_zip }}
-          rm ${{ env.ort_zip }}
+          ls -l
+          ls -R ${{ env.ORT_PACKAGE_NAME }}
+        continue-on-error: true
 
-      - name: Rename OnnxRuntime to ort
+# TODO: Find out why do we need to to have libonnxruntime.so.$ort_version
+      - name: Extra OnnxRuntime library and header files
         run: |
-          mv ${{ env.ort_dir }} ort
+          mkdir -p ort/lib
+          mv ${{ env.ORT_PACKAGE_NAME }}/buildTransitive/native/include ort/
+          mv ${{ env.ORT_PACKAGE_NAME }}/runtimes/linux-x64/native/* ort/lib/
+          ort_version=$(echo ${{ env.ORT_NIGHTLY_VERSION }} | cut -d- -f1-1)
+          cp ort/lib/libonnxruntime.so ort/lib/libonnxruntime.so.$ort_version
+
 
       - name: Get Docker Image
         run: |
@@ -78,7 +110,7 @@ jobs:
             --volume $GITHUB_WORKSPACE:/ort_genai_src \
             -w /ort_genai_src onnxruntimecudabuildx64 \
             bash -c " \
-              /usr/bin/cmake --build --preset linux_gcc_cuda_release --parallel $( nproc )"
+              /usr/bin/cmake --build --preset linux_gcc_cuda_release"
 
       - name: Get HuggingFace Token
         run: |

.github/workflows/mac-cpu-arm64-build.yml

@@ -4,9 +4,8 @@ concurrency:
   group: ${{ github.workflow }}-${{ github.head_ref || github.run_id }}
   cancel-in-progress: true
 env:
-  ort_dir: "onnxruntime-osx-arm64-1.17.3"
-  ort_zip: "onnxruntime-osx-arm64-1.17.3.tgz"
-  ort_url: "https://github.com/microsoft/onnxruntime/releases/download/v1.17.3/onnxruntime-osx-arm64-1.17.3.tgz"
+  ORT_NIGHTLY_REST_API: "https://feeds.dev.azure.com/aiinfra/PublicPackages/_apis/packaging/Feeds/ORT-Nightly/packages?packageNameQuery=Microsoft.ML.OnnxRuntime&api-version=6.0-preview.1"
+  ORT_PACKAGE_NAME: "Microsoft.ML.OnnxRuntime"
 jobs:
   mac-cpu-arm64-build:
     runs-on: macos-latest
@@ -16,22 +15,21 @@ jobs:
         with:
           submodules: true
 
-      - name: Install ninja
+      - name: Get the Latest OnnxRuntime Nightly Version
         run: |
-          brew install ninja
-
-      - name: Download OnnxRuntime
+          ORT_NIGHTLY_VERSION=$(curl -s "${{ env.ORT_NIGHTLY_REST_API }}" | jq -r '.value[0].versions[0].normalizedVersion')
+          echo "$ORT_NIGHTLY_VERSION" 
+          echo "ORT_NIGHTLY_VERSION=$ORT_NIGHTLY_VERSION" >> $GITHUB_ENV
+      - name: Download OnnxRuntime Nightly
         run: |
-          curl -L -o ${{ env.ort_zip }} ${{ env.ort_url }} 
+          nuget install ${{ env.ORT_PACKAGE_NAME }} -version ${{ env.ORT_NIGHTLY_VERSION }} -x
 
-      - name: Unzip OnnxRuntime
+      - name: Extra OnnxRuntime library and header files
         run: |
-          tar -xzf ${{ env.ort_zip }}
-          rm ${{ env.ort_zip }}
+          mkdir -p ort/lib
+          mv ${{ env.ORT_PACKAGE_NAME }}/build/native/include ort/
+          mv ${{ env.ORT_PACKAGE_NAME }}/runtimes/osx-arm64/native/* ort/lib/
 
-      - name: Rename OnnxRuntime to ort
-        run: |
-          mv ${{ env.ort_dir }} ort
 
       - name: Configure CMake
         run: |

.github/workflows/win-cpu-arm64-build.yml

@@ -53,6 +53,11 @@ jobs:
       run: |
         cmake --build --preset windows_arm64_cpu_release --parallel
 
+    - name: Build the C# API and Run the C# Tests
+      run: |
+        cd test\csharp
+        dotnet test /p:NativeBuildOutputDir="$env:GITHUB_WORKSPACE\$env:binaryDir\Release"
+
     - name: Install the Python Wheel and Test Dependencies
       run: |
         python -m pip install (Get-ChildItem ("$env:binaryDir\wheel\*.whl"))
@@ -62,10 +67,7 @@ jobs:
       run: |
         python test/python/test_onnxruntime_genai.py --cwd "test\python" --test_models "test\test_models"
 
-    - name: Build the C# API and Run the C# Tests
-      run: |
-        cd test\csharp
-        dotnet test /p:NativeBuildOutputDir="$env:GITHUB_WORKSPACE\$env:binaryDir\Release"
+
 
     - name: Verify Build Artifacts
       if: always()

.github/workflows/win-cpu-x64-build.yml

@@ -11,10 +11,9 @@ concurrency:
   group: ${{ github.workflow }}-${{ github.head_ref || github.run_id }}
   cancel-in-progress: true
 env:
-  ort_dir: "onnxruntime-win-x64-1.17.3"
-  ort_zip: "$(ort_dir).zip"
-  ort_url: "https://github.com/microsoft/onnxruntime/releases/download/v1.17.3/$(ort_zip)"
   binaryDir: 'build/cpu'
+  ORT_NIGHTLY_REST_API: "https://feeds.dev.azure.com/aiinfra/PublicPackages/_apis/packaging/Feeds/ORT-Nightly/packages?packageNameQuery=Microsoft.ML.OnnxRuntime&api-version=6.0-preview.1"
+  ORT_PACKAGE_NAME: "Microsoft.ML.OnnxRuntime"
 
 jobs:
   windows-cpu-x64-build:
@@ -33,19 +32,32 @@ jobs:
       with:
         vs-version: '17.5'
 
-    - name: Download OnnxRuntime
+    - uses: actions/setup-dotnet@v4
+      with:
+        dotnet-version: '6.0.x'
+
+    - name : Install jq and nuget
       run: |
-        $env:ort_url = "https://github.com/microsoft/onnxruntime/releases/download/v1.17.3/onnxruntime-win-x64-1.17.3.zip"
-        Invoke-WebRequest -Uri $env:ort_url -OutFile $env:ort_zip
+        choco install -y jq curl
 
-    - name: Unzip OnnxRuntime
+    - name: Get the Latest OnnxRuntime Nightly Version
+      shell: pwsh
       run: |
-        Expand-Archive $env:ort_zip -DestinationPath .
-        Remove-Item -Path $env:ort_zip
+        $ORT_NIGHTLY_VERSION = $(curl -s "${{ env.ORT_NIGHTLY_REST_API }}" | jq -r '.value[0].versions[0].normalizedVersion')
+        echo "$ORT_NIGHTLY_VERSION" 
+        "ORT_NIGHTLY_VERSION=$ORT_NIGHTLY_VERSION" | Out-File -FilePath $env:GITHUB_ENV -Append
+    - name: Download OnnxRuntime Nightly
+      run: |
+        nuget install ${{ env.ORT_PACKAGE_NAME }} -version ${{ env.ORT_NIGHTLY_VERSION }} -x -NonInteractive
+
+    - run: Get-ChildItem  ${{ env.ORT_PACKAGE_NAME }} -Recurse
+      continue-on-error: true
 
-    - name: Rename OnnxRuntime to ort
+    - name: Extra OnnxRuntime library and header files
       run: |
-        Rename-Item -Path $env:ort_dir -NewName ort
+        mkdir ort/lib
+        move ${{ env.ORT_PACKAGE_NAME }}/build/native/include ort/
+        move ${{ env.ORT_PACKAGE_NAME }}/runtimes/win-x64/native/* ort/lib/
 
     - name: Initialize CodeQL
       uses: github/codeql-action/init@v3
@@ -60,6 +72,11 @@ jobs:
       run: |
         cmake --build --preset windows_x64_cpu_release --parallel
 
+    - name: Build the C# API and Run the C# Tests
+      run: |
+        cd test\csharp
+        dotnet test /p:NativeBuildOutputDir="$env:GITHUB_WORKSPACE\$env:binaryDir\Release"
+
     - name: Install the python wheel and test dependencies
       run: |
         python -m pip install (Get-ChildItem ("$env:binaryDir\wheel\*.whl"))
@@ -76,10 +93,7 @@ jobs:
       run: |
         python test/python/test_onnxruntime_genai.py --cwd "test\python" --test_models "test\test_models"
 
-    - name: Build the C# API and Run the C# Tests
-      run: |
-        cd test\csharp
-        dotnet test /p:NativeBuildOutputDir="$env:GITHUB_WORKSPACE\$env:binaryDir\Release"
+
 
     - name: Verify Build Artifacts
       if: always()

.github/workflows/win-cuda-x64-build.yml

@@ -8,14 +8,12 @@ concurrency:
 env:
   AZCOPY_AUTO_LOGIN_TYPE: MSI
   AZCOPY_MSI_CLIENT_ID: 63b63039-6328-442f-954b-5a64d124e5b4
-  ort_dir: "onnxruntime-win-x64-gpu-1.17.3"
-  ort_zip: "onnxruntime-win-x64-gpu-1.17.3.zip"
-  ort_url: "https://github.com/microsoft/onnxruntime/releases/download/v1.17.3/onnxruntime-win-x64-gpu-1.17.3.zip"
   cuda_dir: "${{ github.workspace }}\\cuda_sdk"
   cuda_version: "11.8"
   CUDA_PATH: ${{ github.workspace }}\\cuda_sdk\\v11.8
   binaryDir: 'build/cuda'
-
+  ORT_NIGHTLY_REST_API: "https://feeds.dev.azure.com/aiinfra/PublicPackages/_apis/packaging/Feeds/ORT-Nightly/packages?packageNameQuery=Microsoft.ML.OnnxRuntime.Gpu.Windows&api-version=6.0-preview.1"
+  ORT_PACKAGE_NAME: "Microsoft.ML.OnnxRuntime.Gpu.Windows"
 
 jobs:
   windows-cuda-x64-build:
@@ -35,17 +33,32 @@ jobs:
       run: |
         azcopy.exe cp --recursive "https://lotusscus.blob.core.windows.net/models/cuda_sdk/v${{ env.cuda_version }}" ${{ env.cuda_dir}}
 
-    - name: Download OnnxRuntime
+    - uses: actions/setup-dotnet@v4
+      with:
+        dotnet-version: '6.0.x'
+
+    - name : Install jq and curl
       run: |
-        Invoke-WebRequest -Uri $env:ort_url -OutFile $env:ort_zip
+        choco install -y jq curl
 
-    - name: Unzip OnnxRuntime
+    - name: Get the Latest OnnxRuntime Nightly Version
+      shell: pwsh
+      run: |
+        $ORT_NIGHTLY_VERSION=$(curl -s "${{ env.ORT_NIGHTLY_REST_API }}" | jq -r '.value[0].versions[0].normalizedVersion')
+        echo "$ORT_NIGHTLY_VERSION" 
+        "ORT_NIGHTLY_VERSION=$ORT_NIGHTLY_VERSION" | Out-File -FilePath $env:GITHUB_ENV -Append
+    - name: Download OnnxRuntime Nightly
       run: |
-        Expand-Archive $env:ort_zip -DestinationPath .
-        Remove-Item -Path $env:ort_zip
-    - name: Rename OnnxRuntime to ort
+        nuget install ${{ env.ORT_PACKAGE_NAME }} -version ${{ env.ORT_NIGHTLY_VERSION }} -ExcludeVersion -NonInteractive
+
+    - run: Get-ChildItem  ${{ env.ORT_PACKAGE_NAME }} -Recurse
+      continue-on-error: true
+
+    - name: Extra OnnxRuntime library and header files
       run: |
-        Rename-Item -Path $env:ort_dir -NewName ort
+        mkdir ort/lib
+        move ${{ env.ORT_PACKAGE_NAME }}/buildTransitive/native/include ort/
+        move ${{ env.ORT_PACKAGE_NAME }}/runtimes/win-x64/native/* ort/lib/    
 
     - name: Configure CMake
       run: |
@@ -59,6 +72,11 @@ jobs:
       run: |
         echo "${{ env.cuda_dir }}\\v${{ env.cuda_version }}\\bin" | Out-File -FilePath $env:GITHUB_PATH -Encoding utf8 -Append
 
+    - name: Build the C# API and Run the C# Tests
+      run: |
+        cd test\csharp
+        dotnet test /p:Configuration=release /p:NativeBuildOutputDir="$env:GITHUB_WORKSPACE\$env:binaryDir\Release"
+
     - name: Install the Python Wheel and Test Dependencies
       run: |
         python -m pip install (Get-ChildItem ("$env:binaryDir\wheel\*.whl"))
@@ -75,10 +93,6 @@ jobs:
       run: |
         python test/python/test_onnxruntime_genai.py --cwd "test\python" --test_models "test\test_models"
 
-    - name: Build the C# API and Run the C# Tests
-      run: |
-        cd test\csharp
-        dotnet test /p:Configuration=release /p:NativeBuildOutputDir="$env:GITHUB_WORKSPACE\$env:binaryDir\Release"
 
     - name: Verify Build Artifacts
       if: always()

cmake/presets/CMakeMacOSConfigPresets.json

@@ -6,7 +6,7 @@
   "configurePresets": [
         {
       "name": "macos_default",
-      "generator": "Ninja",
+      "generator": "Unix Makefiles",
       "binaryDir": "${sourceDir}/build/cpu",
       "cacheVariables": {
         "CMAKE_POSITION_INDEPENDENT_CODE": "ON",