[TensorRT EP] New CIs to test TRT+minimal CUDA build (#23028)

### Description
<!-- Describe your changes. -->
New CI:
[Linux_TRT_Minimal_CUDA_Test_CI](https://dev.azure.com/onnxruntime/onnxruntime/_build?definitionId=230&_a=summary)
and [Win_TRT_Minimal_CUDA_Test_CI
](https://dev.azure.com/onnxruntime/onnxruntime/_build?definitionId=231)
Setting config for new CI to monitor if there's no issue to build
ORT-TRTEP with minimal CUDA
* yaml content is following Linux TRT CI yaml, with different build
arg/cache name
* build arg is following [[TensorRT EP] Enable a minimal CUDA EP
compilation without
kernels](https://github.com/microsoft/onnxruntime/pull/19052#issuecomment-1888066851)



### Motivation and Context
<!-- - Why is this change required? What problem does it solve?
- If it fixes an open issue, please link to the issue here. -->
Monitor if user is able to build ORT-TRTEP-minimalCUDA without any
blocker
(which takes ~30min to build)

tools/ci_build/build.py

@@ -260,6 +260,7 @@ def parse_arguments():
     )
 
     parser.add_argument("--disable_cuda_nhwc_ops", action="store_true", help="Disable CUDA NHWC ops in build.")
+    parser.add_argument("--enable_cuda_minimal_build", action="store_true", help="Enable CUDA minimal build.")
 
     # Python bindings
     parser.add_argument("--enable_pybind", action="store_true", help="Enable Python Bindings.")
@@ -1093,6 +1094,7 @@ def generate_build_tree(
         "-Donnxruntime_DISABLE_FLOAT8_TYPES=" + ("ON" if disable_float8_types else "OFF"),
         "-Donnxruntime_DISABLE_SPARSE_TENSORS=" + ("ON" if disable_sparse_tensors else "OFF"),
         "-Donnxruntime_DISABLE_OPTIONAL_TYPE=" + ("ON" if disable_optional_type else "OFF"),
+        "-Donnxruntime_CUDA_MINIMAL=" + ("ON" if args.enable_cuda_minimal_build else "OFF"),
     ]
 
     if args.rv64:

tools/ci_build/github/azure-pipelines/linux-gpu-tensorrt-cuda-minimal-ci-pipeline.yml

@@ -0,0 +1,108 @@
+##### start trigger Don't edit it manually, Please do edit set-trigger-rules.py ####
+### please do rerun set-trigger-rules.py ###
+trigger:
+  branches:
+    include:
+    - main
+    - rel-*
+  paths:
+    exclude:
+    - docs/**
+    - README.md
+    - CONTRIBUTING.md
+    - BUILD.md
+    - 'js/web'
+    - 'onnxruntime/core/providers/js'
+pr:
+  branches:
+    include:
+    - main
+    - rel-*
+  paths:
+    exclude:
+    - docs/**
+    - README.md
+    - CONTRIBUTING.md
+    - BUILD.md
+    - 'js/web'
+    - 'onnxruntime/core/providers/js'
+#### end trigger ####
+parameters:
+  - name: CudaVersion
+    displayName: CUDA version
+    type: string
+    default: '12.2'
+    values:
+      - 11.8
+      - 12.2
+
+variables:
+  - template: templates/common-variables.yml
+  - name: docker_base_image
+    ${{ if eq(parameters.CudaVersion, '11.8') }}:
+      value: onnxruntimebuildcache.azurecr.io/internal/azureml/onnxruntime/build/cuda11_x64_almalinux8_gcc11:20241120.3
+    ${{ if eq(parameters.CudaVersion, '12.2') }}:
+      value: onnxruntimebuildcache.azurecr.io/internal/azureml/onnxruntime/build/cuda12_x64_ubi8_gcc12:20241120.3
+  - name: linux_trt_version
+    ${{ if eq(parameters.CudaVersion, '11.8') }}:
+      value: ${{ variables.linux_trt_version_cuda11 }}
+    ${{ if eq(parameters.CudaVersion, '12.2') }}:
+      value: ${{ variables.linux_trt_version_cuda12 }}
+
+jobs:
+- job: Linux_Build
+  timeoutInMinutes: 180
+  variables:
+    skipComponentGovernanceDetection: true
+    ALLOW_RELEASED_ONNX_OPSET_ONLY: '1'
+    ORT_CACHE_DIR: '$(Agent.TempDirectory)/ort/ccache'
+    TODAY: $[format('{0:dd}{0:MM}{0:yyyy}', pipeline.startTime)]
+  workspace:
+    clean: all
+  pool: onnxruntime-tensorrt-linuxbuild-T4
+  steps:
+  - task: mspremier.PostBuildCleanup.PostBuildCleanup-task.PostBuildCleanup@3
+    displayName: 'Clean Agent Directories'
+    condition: always()
+
+  - checkout: self
+    clean: true
+    submodules: none
+
+  - template: templates/get-docker-image-steps.yml
+    parameters:
+      Dockerfile: tools/ci_build/github/linux/docker/Dockerfile.manylinux2_28_cuda
+      Context: tools/ci_build/github/linux/docker
+      DockerBuildArgs: "
+      --network=host
+      --build-arg BASEIMAGE=${{ variables.docker_base_image }}
+      --build-arg TRT_VERSION=${{ variables.linux_trt_version }}
+      --build-arg BUILD_UID=$( id -u )
+      "
+      Repository: onnxruntimetensorrtcudaminimalbuild
+
+  - template: templates/linux-build-step-with-cache.yml
+    parameters:
+      WithCache: true
+      Today: $(TODAY)
+      AdditionalKey: gpu_tensorrt_cuda_minimal
+      CacheDir: '$(ORT_CACHE_DIR)'
+      BuildStep:
+        - task: CmdLine@2
+          inputs:
+            script: |
+              docker run --gpus all --rm \
+                  --volume /data/onnx:/data/onnx:ro \
+                  --volume $(Build.SourcesDirectory):/onnxruntime_src \
+                  --volume $(Build.BinariesDirectory):/build \
+                  --volume /data/models:/build/models:ro \
+                  --volume $HOME/.onnx:/home/onnxruntimedev/.onnx \
+                  --volume $(ORT_CACHE_DIR):/cache \
+                  -e ALLOW_RELEASED_ONNX_OPSET_ONLY=0 \
+                  -e NIGHTLY_BUILD \
+                  -e BUILD_BUILDNUMBER \
+                  -e CCACHE_DIR=/cache -w /onnxruntime_src \
+                  onnxruntimetensorrtcudaminimalbuild tools/ci_build/github/linux/build_tensorrt_ci.sh --cuda_minimal=ON
+            workingDirectory: $(Build.SourcesDirectory)
+
+  - template: templates/explicitly-defined-final-tasks.yml

tools/ci_build/github/azure-pipelines/win-gpu-tensorrt-cuda-minimal-ci-pipeline.yml

@@ -0,0 +1,86 @@
+##### start trigger Don't edit it manually, Please do edit set-trigger-rules.py ####
+### please do rerun set-trigger-rules.py ###
+trigger:
+  branches:
+    include:
+    - main
+    - rel-*
+  paths:
+    exclude:
+    - docs/**
+    - README.md
+    - CONTRIBUTING.md
+    - BUILD.md
+    - 'js/web'
+    - 'onnxruntime/core/providers/js'
+pr:
+  branches:
+    include:
+    - main
+    - rel-*
+  paths:
+    exclude:
+    - docs/**
+    - README.md
+    - CONTRIBUTING.md
+    - BUILD.md
+    - 'js/web'
+    - 'onnxruntime/core/providers/js'
+#### end trigger ####
+parameters:
+- name: CudaVersion
+  displayName: CUDA version
+  type: string
+  default: '12.2'
+  values:
+    - 11.8
+    - 12.2
+
+variables:
+  - template: templates/common-variables.yml
+  - name: win_trt_folder
+    ${{ if eq(parameters.CudaVersion, '11.8') }}:
+      value: ${{ variables.win_trt_folder_cuda11 }}
+    ${{ if eq(parameters.CudaVersion, '12.2') }}:
+      value: ${{ variables.win_trt_folder_cuda12 }}
+
+jobs:
+- job: 'build'
+  pool: 'onnxruntime-Win2022-GPU-A10'
+  variables:
+    MsbuildArguments: '-detailedsummary -maxcpucount -consoleloggerparameters:PerformanceSummary'
+    EnvSetupScript: setup_env_trt.bat
+    skipComponentGovernanceDetection: true
+    TODAY: $[format('{0:dd}{0:MM}{0:yyyy}', pipeline.startTime)]
+  timeoutInMinutes: 150
+  workspace:
+    clean: all
+  steps:
+  - template: templates/jobs/win-ci-prebuild-steps.yml
+    parameters:
+      EnvSetupScript: $(EnvSetupScript)
+      DownloadCUDA: true
+      DownloadTRT: true
+      BuildArch: 'x64'
+      BuildConfig: RelWithDebInfo
+      MachinePool: 'onnxruntime-Win2022-GPU-A10'
+      WithCache: true
+      Today: $(Today)
+
+  - template: templates/jobs/win-ci-build-steps.yml
+    parameters:
+      WithCache: True
+      Today: $(TODAY)
+      AdditionalKey: "gpu_tensorrt_cuda_minimal | RelWithDebInfo"
+      BuildPyArguments: '--config RelWithDebInfo --parallel --use_binskim_compliant_compile_flags --build_dir $(Build.BinariesDirectory) --skip_submodule_sync --build_shared_lib --update --cmake_generator "Visual Studio 17 2022" --build_wheel --enable_onnx_tests --use_tensorrt --tensorrt_home="$(Agent.TempDirectory)\${{ variables.win_trt_folder }}" --cuda_home="$(Agent.TempDirectory)\v${{ parameters.CudaVersion }}" --cmake_extra_defines CMAKE_CUDA_ARCHITECTURES=86 --enable_cuda_minimal_build'
+      MsbuildArguments: $(MsbuildArguments)
+      BuildArch: 'x64'
+      Platform: 'x64'
+      BuildConfig: RelWithDebInfo
+
+  - task: PythonScript@0
+    displayName: 'Build wheel'
+    inputs:
+      scriptPath: '$(Build.SourcesDirectory)\setup.py'
+      arguments: 'bdist_wheel'
+      workingDirectory: '$(Build.BinariesDirectory)\RelWithDebInfo\RelWithDebInfo'

tools/ci_build/github/linux/build_tensorrt_ci.sh

@@ -21,6 +21,19 @@ BUILD_ARGS=('--config' 'Release'
 	      "CMAKE_CUDA_ARCHITECTURES=75"
 	      "onnxruntime_BUILD_UNIT_TESTS=ON"
 	      "onnxruntime_ENABLE_CUDA_EP_INTERNAL_TESTS=ON")
+
+# Parse external args
+for arg in "$@"; do
+  case $arg in
+    --cuda_minimal=ON)
+      # Replace onnxruntime_BUILD_UNIT_TESTS=ON with OFF
+      BUILD_ARGS=("${BUILD_ARGS[@]/onnxruntime_BUILD_UNIT_TESTS=ON/onnxruntime_BUILD_UNIT_TESTS=OFF}")
+      BUILD_ARGS+=("--enable_cuda_minimal_build")
+      BUILD_ARGS+=("--skip_tests")
+      ;;
+  esac
+done
+
 if [ -x "$(command -v ninja)" ]; then
     BUILD_ARGS+=('--cmake_generator' 'Ninja')
 fi