Allowing ADO packing pipelines to use the specific ORT commit

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

@@ -58,7 +58,7 @@ jobs:
         continue-on-error: true
 
 # TODO: Find out why do we need to to have libonnxruntime.so.$ort_version
-      - name: Extra OnnxRuntime library and header files
+      - name: Extract OnnxRuntime library and header files
         run: |
           mkdir -p ort/lib
           mv ${{ env.ORT_PACKAGE_NAME }}/build/native/include ort/

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

@@ -72,7 +72,7 @@ jobs:
         continue-on-error: true
 
 # TODO: Find out why do we need to to have libonnxruntime.so.$ort_version
-      - name: Extra OnnxRuntime library and header files
+      - name: Extract OnnxRuntime library and header files
         run: |
           mkdir -p ort/lib
           mv ${{ env.ORT_PACKAGE_NAME }}/buildTransitive/native/include ort/

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

@@ -30,7 +30,7 @@ jobs:
         run: |
           nuget install ${{ env.ORT_PACKAGE_NAME }} -version ${{ env.ORT_NIGHTLY_VERSION }} -x
 
-      - name: Extra OnnxRuntime library and header files
+      - name: Extract OnnxRuntime library and header files
         run: |
           mkdir -p ort/lib
           mv ${{ env.ORT_PACKAGE_NAME }}/build/native/include ort/

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

@@ -53,7 +53,7 @@ jobs:
     - run: Get-ChildItem  ${{ env.ORT_PACKAGE_NAME }} -Recurse
       continue-on-error: true
 
-    - name: Extra OnnxRuntime library and header files
+    - name: Extract OnnxRuntime library and header files
       run: |
         mkdir ort/lib
         move ${{ env.ORT_PACKAGE_NAME }}/build/native/include ort/

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

@@ -60,7 +60,7 @@ jobs:
     - run: Get-ChildItem  ${{ env.ORT_PACKAGE_NAME }} -Recurse
       continue-on-error: true
 
-    - name: Extra OnnxRuntime library and header files
+    - name: Extract OnnxRuntime library and header files
       run: |
         mkdir ort/lib
         move ${{ env.ORT_PACKAGE_NAME }}/buildTransitive/native/include ort/

.pipelines/nuget-publishing.yml

@@ -27,12 +27,17 @@ parameters:
 - name: ort_version
   displayName: 'OnnxRuntime version'
   type: string
-  default: '1.17.3'
+  default: '1.18.0-dev-20240426-1256-b842effa29'
+
+- name: ort_cuda_version
+  displayName: 'OnnxRuntime GPU version'
+  type: string
+  default: '1.18.0-dev-20240426-0614-b842effa29'
 
 - name: ort_dml_version
-  displayName: 'OnnxRuntime DirectML version'
+  displayName: 'OnnxRuntime DML version'
   type: string
-  default: '1.18.0-dev-20240423-0527-c07b8d545d'
+  default: '1.18.0-dev-20240426-0116-b842effa29'
 
 - name: cuda_version
   displayName: 'CUDA version'
@@ -42,11 +47,6 @@ parameters:
   - '12.2'
   default: '11.8'
 
-- name: publish_to_ado_feed
-  displayName: 'Publish to Azure DevOps Feed'
-  type: boolean
-  default: false
-
 resources:
   repositories:
   - repository: manylinux
@@ -65,7 +65,6 @@ stages:
     enable_linux_cuda: ${{ parameters.enable_linux_cuda }}
     enable_win_dml: ${{ parameters.enable_win_dml }}
     ort_version: ${{ parameters.ort_version }}
+    ort_cuda_version: ${{ parameters.ort_cuda_version }}
     ort_dml_version: ${{ parameters.ort_dml_version }}
     cuda_version: ${{ parameters.cuda_version }}
-    publish_to_ado_feed: ${{ parameters.publish_to_ado_feed }}
-

.pipelines/pypl-publishing.yml

@@ -7,7 +7,7 @@ parameters:
 - name: enable_win_cuda
   displayName: 'Whether Windows CUDA package is built.'
   type: boolean
-  default : true
+  default: true
 
 - name: enable_win_dml
   displayName: 'Whether Windows DirectML package is built.'
@@ -27,12 +27,17 @@ parameters:
 - name: ort_version
   displayName: 'OnnxRuntime version'
   type: string
-  default: '1.17.3'
+  default: '1.18.0-dev-20240426-1256-b842effa29'
+
+- name: ort_cuda_version
+  displayName: 'OnnxRuntime GPU version'
+  type: string
+  default: '1.18.0-dev-20240426-0614-b842effa29'
 
 - name: ort_dml_version
-  displayName: 'OnnxRuntime DirectML version'
+  displayName: 'OnnxRuntime DML version'
   type: string
-  default: '1.18.0-dev-20240423-0527-c07b8d545d'
+  default: '1.18.0-dev-20240426-0116-b842effa29'
 
 - name: cuda_version
   displayName: 'CUDA version'
@@ -42,11 +47,6 @@ parameters:
   - '11.8'
   - '12.2'
 
-- name: publish_to_ado_feed
-  displayName: 'Whether to publish the packages to ADO feed.'
-  type: boolean
-  default: false
-
 resources:
   repositories:
   - repository: manylinux
@@ -65,6 +65,7 @@ stages:
     enable_win_cuda: ${{ parameters.enable_win_cuda }}
     enable_win_dml: ${{ parameters.enable_win_dml }}
     ort_version: ${{ parameters.ort_version }}
+    ort_cuda_version: ${{ parameters.ort_cuda_version }}
     ort_dml_version: ${{ parameters.ort_dml_version }}
     cuda_version: ${{ parameters.cuda_version }}
-    publish_to_ado_feed: ${{ parameters.publish_to_ado_feed }}
+

.pipelines/stages/jobs/nuget-packaging-job.yml

@@ -13,8 +13,6 @@ parameters:
   values:
   - 'linux'
   - 'win'
-- name: publish_to_ado_feed
-  type: boolean
 
 jobs:
 - job: nuget_${{ parameters.os }}_${{ parameters.ep }}_${{ parameters.arch }}_packaging
@@ -44,18 +42,21 @@ jobs:
     value: ${{ parameters.ort_version }}
   - name: GDN_CODESIGN_TARGETDIRECTORY
     value: '$(Build.ArtifactStagingDirectory)/nuget'
+  - name: os
+    value: ${{ parameters.os }}
   - name: ort_filename
     ${{ if eq(parameters.ep, 'cpu') }}:
-      value: 'onnxruntime-${{ parameters.os }}-${{ parameters.arch }}-${{ parameters.ort_version }}'
+      value: 'Microsoft.ML.OnnxRuntime'
     ${{ elseif eq(parameters.ep, 'cuda') }}:
-      ${{if eq(parameters.cuda_version, '11.8') }}:
-        value: 'onnxruntime-${{ parameters.os }}-${{ parameters.arch }}-gpu-${{ parameters.ort_version }}'
+      ${{if eq(parameters.os, 'win') }}:
+        value: 'Microsoft.ML.OnnxRuntime.Gpu.Windows'
       ${{ else }}:
-        value: 'onnxruntime-${{ parameters.os }}-${{ parameters.arch }}-cuda12-${{ parameters.ort_version }}'
+        value: 'Microsoft.ML.OnnxRuntime.Gpu.Linux'
     ${{ elseif eq(parameters.ep, 'directml')}}:
-      value: 'Microsoft.ML.OnnxRuntime.DirectML.${{ parameters.ort_version }}'
+      value: 'Microsoft.ML.OnnxRuntime.DirectML'
     ${{ else }}:
-      value: 'onnxruntime-${{ parameters.os }}-${{ parameters.arch }}-${{ parameters.ep}}-${{ parameters.ort_version }}'
+      value: 'Microsoft.ML.OnnxRuntime'
+
   - name: genai_nuget_ext
     ${{ if eq(parameters.ep, 'cpu') }}:
       value: ''
@@ -84,15 +85,18 @@ jobs:
     - template: steps/capi-linux-step.yml
       parameters:
         target: 'onnxruntime-genai'
+        arch: ${{ parameters.arch }}
+        ep: ${{ parameters.ep }}
+
     # TODO: Add a step to build the linux nuget package
 
   - ${{ if eq(parameters.os, 'win') }}:
     - template: steps/capi-win-step.yml
       parameters:
         target: 'onnxruntime-genai'
+        arch: ${{ parameters.arch }}
+        ep: ${{ parameters.ep }}
     - template: steps/nuget-win-step.yml
-    - ${{ if eq(parameters.publish_to_ado_feed, true)}}:
-      - template: steps/nuget-ado-feed-releasing-step.yml
 
   - template: steps/compliant-and-cleanup-step.yml
 

.pipelines/stages/jobs/py-packaging-job.yml

@@ -13,8 +13,7 @@ parameters:
   values:
   - 'linux'
   - 'win'
-- name: publish_to_ado_feed
-  type: boolean
+
 
 jobs:
 - job: python_${{ parameters.os }}_${{ parameters.ep }}_${{ parameters.arch }}_packaging
@@ -67,18 +66,21 @@ jobs:
     value: ${{ parameters.ep }}
   - name: ort_version
     value: ${{ parameters.ort_version }}
+  - name: os
+    value: ${{ parameters.os }}
+
   - name: ort_filename
     ${{ if eq(parameters.ep, 'cpu') }}:
-      value: 'onnxruntime-${{ parameters.os }}-${{ parameters.arch }}-${{ parameters.ort_version }}'
+      value: 'Microsoft.ML.OnnxRuntime'
     ${{ elseif eq(parameters.ep, 'cuda') }}:
-      ${{if eq(parameters.cuda_version, '11.8') }}:
-        value: 'onnxruntime-${{ parameters.os }}-${{ parameters.arch }}-gpu-${{ parameters.ort_version }}'
+      ${{if eq(parameters.os, 'win') }}:
+        value: 'Microsoft.ML.OnnxRuntime.Gpu.Windows'
       ${{ else }}:
-        value: 'onnxruntime-${{ parameters.os }}-${{ parameters.arch }}-cuda12-${{ parameters.ort_version }}'
+        value: 'Microsoft.ML.OnnxRuntime.Gpu.Linux'
     ${{ elseif eq(parameters.ep, 'directml')}}:
-      value: 'Microsoft.ML.OnnxRuntime.DirectML.${{ parameters.ort_version }}'
+      value: 'Microsoft.ML.OnnxRuntime.DirectML'
     ${{ else }}:
-      value: 'onnxruntime-${{ parameters.os }}-${{ parameters.arch }}-${{ parameters.ep}}-${{ parameters.ort_version }}'
+      value: 'Microsoft.ML.OnnxRuntime'
 
   - name: dml_dir
     value: 'Microsoft.AI.DirectML.1.14.1'
@@ -107,16 +109,16 @@ jobs:
     - template: steps/capi-linux-step.yml
       parameters:
         target: 'python'
+        arch: ${{ parameters.arch }}
+        ep: ${{ parameters.ep }}
 
   # Windows job needs to set the python version and install the required packages
   - ${{ if eq(parameters.os, 'win') }}:
     - template: steps/capi-win-step.yml
       parameters:
         target: 'python'
+        arch: ${{ parameters.arch }}
         ep: ${{ parameters.ep }}
 
-  - ${{ if eq(parameters.publish_to_ado_feed, true)}}:
-    - template: steps/py-ado-feed-releasing-step.yml
-
   - template: steps/compliant-and-cleanup-step.yml
 

.pipelines/stages/jobs/steps/capi-linux-step.yml

@@ -1,6 +1,12 @@
 parameters:
 - name: target
   type: string
+- name: ep
+  type: string
+  default: 'cpu'
+- name: arch
+  type: string
+  default: 'x64'
 steps:
 
 - checkout: self
@@ -26,6 +32,7 @@ steps:
     echo "arch=$(arch)"
     echo "ep=$(ep)"
   displayName: 'Print Parameters'
+
 - template: utils/download-ort.yml
   parameters:
     archiveType: 'tgz'
@@ -40,7 +47,7 @@ steps:
       --container-registry onnxruntimebuildcache \
       --manylinux-src manylinux \
       --multiple_repos \
-      --repository onnxruntime$(ep)build$(arch)
+      --repository ortgenai$(ep)build$(arch)
   displayName: 'Get Docker Image'
   workingDirectory: '$(Build.Repository.LocalPath)'
 
@@ -50,14 +57,15 @@ steps:
       docker run \
       --rm \
       --volume $(Build.Repository.LocalPath):/ort_genai_src \
-      -w /ort_genai_src/ onnxruntime$(ep)build$(arch) \
+      -w /ort_genai_src/ ortgenai$(ep)build$(arch) \
       bash -c " \
           /usr/bin/cmake --preset linux_gcc_$(ep)_release \
             -DENABLE_TESTS=OFF && \
           /usr/bin/cmake --build --preset linux_gcc_$(ep)_release \
             --target onnxruntime-genai"
     displayName: 'Build GenAi'
     workingDirectory: '$(Build.Repository.LocalPath)'
+
   - task: BinSkim@4
     displayName: 'Run BinSkim'
     inputs:
@@ -66,20 +74,30 @@ steps:
   - template: utils/capi-archive.yml
     parameters:
       archiveType: tar
+  - script: |
+      set -e -x
+      docker run \
+      --rm \
+      --volume $(Build.Repository.LocalPath):/ort_genai_src \
+      -w /ort_genai_src/ ortgenai$(ep)build$(arch) \
+      bash -c " \
+          /usr/bin/cmake --build --preset linux_gcc_$(ep)_release --target package"
+    displayName: 'Package C/C++ API'
+    workingDirectory: '$(Build.Repository.LocalPath)'
 
   - task: PublishBuildArtifacts@1
     displayName: 'Publish Artifact: ONNXRuntime Genai capi'
     inputs:
       ArtifactName: $(artifactName)-capi
-      PathtoPublish: '$(Build.ArtifactStagingDirectory)/capi'
+      PathtoPublish: '$(Build.Repository.LocalPath)/build/$(ep)/package'
 
 - ${{ if eq(parameters.target, 'python') }}:
   - bash: |
       set -e -x
       docker run \
       --rm \
       --volume $(Build.Repository.LocalPath):/ort_genai_src \
-      -w /ort_genai_src/ onnxruntime$(ep)build$(arch) \
+      -w /ort_genai_src/ ortgenai$(ep)build$(arch) \
       bash -c " \
           /usr/bin/cmake --preset linux_gcc_$(ep)_release \
             -DENABLE_TESTS=OFF \
@@ -101,7 +119,7 @@ steps:
       docker run \
       --rm \
       --volume $(Build.Repository.LocalPath):/ort_genai_src \
-      -w /ort_genai_src/ onnxruntime$(ep)build$(arch) \
+      -w /ort_genai_src/ ortgenai$(ep)build$(arch) \
       bash -c " \
           /usr/bin/cmake --build --preset linux_gcc_$(ep)_release \
             -DENABLE_TESTS=OFF \

.pipelines/stages/jobs/steps/capi-win-step.yml

@@ -5,6 +5,9 @@ parameters:
 - name: ep
   type: string
   default: 'cpu'
+- name: arch
+  type: string
+  default: 'x64'
 steps:
 - bash: |
     echo "##[error]Error: ep and arch are not set"
@@ -31,6 +34,8 @@ steps:
     echo "cuda_version=$(cuda_version)"
     echo "target=${{ parameters.target }}"
   displayName: 'Print Parameters'
+
+
 - template: utils/download-ort.yml
   parameters:
     archiveType: 'zip'
@@ -83,15 +88,16 @@ steps:
       AnalyzeTargetGlob: '$(Build.Repository.LocalPath)\**\*genai.dll'
     continueOnError: true
 
-  - template: utils/capi-archive.yml
-    parameters:
-      archiveType: zip
+  - powershell: |
+      cmake --build --preset windows_$(arch)_$(ep)_release --target package
+    displayName: 'Package C/C++ API'
+    workingDirectory: '$(Build.Repository.LocalPath)'
 
   - task: PublishBuildArtifacts@1
     displayName: 'Publish Artifact: ONNXRuntime Genai capi'
     inputs:
       ArtifactName: $(artifactName)-capi
-      PathtoPublish: '$(Build.ArtifactStagingDirectory)/capi'
+      PathtoPublish: '$(Build.Repository.LocalPath)\build\$(ep)\package'
 
 - ${{ if eq(parameters.target, 'python') }}:
   - task: BinSkim@4