[QNN EP] Reshape Transpose pattern fusion for 6D tensors

.github/workflows/gradle-wrapper-validation.yml

@@ -16,7 +16,7 @@ jobs:
     runs-on: ["self-hosted", "1ES.Pool=onnxruntime-github-Ubuntu2204-AMD-CPU"]
     steps:
       - uses: actions/checkout@v5
-      - uses: gradle/actions/wrapper-validation@v4
+      - uses: gradle/actions/wrapper-validation@v5
 concurrency:
   group: ${{ github.workflow }}-${{ github.event_name == 'pull_request' && github.ref || github.sha }}
   cancel-in-progress: true

.github/workflows/publish-csharp-apidocs.yml

@@ -20,7 +20,7 @@ permissions:
 
 jobs:
   build:
-    runs-on: ["self-hosted", "1ES.Pool=onnxruntime-github-vs2022-mms"]
+    runs-on: ["self-hosted", "1ES.Pool=onnxruntime-github-vs2022-latest"]
     env:
       DOCFXVERSION: 2.62.2
     steps:

.github/workflows/windows_cuda.yml

@@ -19,7 +19,7 @@ concurrency:
 jobs:
   build:
     name: Windows GPU CUDA CI Pipeline
-    runs-on: ["self-hosted", "1ES.Pool=onnxruntime-github-vs2022-mms"]
+    runs-on: ["self-hosted", "1ES.Pool=onnxruntime-github-vs2022-latest"]
     steps:
       - uses: actions/checkout@v5
         with:
@@ -41,20 +41,20 @@ jobs:
         working-directory: ${{ github.workspace }}
         shell: cmd
 
-      - name: Download CUDA SDK v12.2
+      - name: Download CUDA SDK v12.8
         working-directory: ${{ runner.temp }}
         run: |
-          azcopy.exe cp --recursive "https://lotusscus.blob.core.windows.net/models/cuda_sdk/v12.2" .
+          azcopy.exe cp --recursive "https://lotusscus.blob.core.windows.net/models/cuda_sdk/v12.8" .
           dir
         shell: pwsh
 
       - name: Add CUDA to PATH
         shell: powershell
         run: |
           Write-Host "Adding CUDA to PATH"
-          Write-Host "CUDA Path: $env:RUNNER_TEMP\v12.2\bin"
-          Add-Content -Path $env:GITHUB_PATH -Value "$env:RUNNER_TEMP\v12.2\bin"
-          Add-Content -Path $env:GITHUB_PATH -Value "$env:RUNNER_TEMP\v12.2\extras\CUPTI\lib64"
+          Write-Host "CUDA Path: $env:RUNNER_TEMP\v12.8\bin"
+          Add-Content -Path $env:GITHUB_PATH -Value "$env:RUNNER_TEMP\v12.8\bin"
+          Add-Content -Path $env:GITHUB_PATH -Value "$env:RUNNER_TEMP\v12.8\extras\CUPTI\lib64"
 
       - uses: actions/setup-node@v5
         with:
@@ -111,7 +111,7 @@ jobs:
             exit $lastExitCode
           }
           # Execute the build process
-          python.exe ${{ github.workspace }}\tools\ci_build\build.py --update --build --config RelWithDebInfo --build_dir build --skip_submodule_sync --build_csharp --parallel --use_binskim_compliant_compile_flags --cmake_generator "Visual Studio 17 2022" --build_shared_lib --build_wheel --build_java --use_cuda --cuda_home="$env:RUNNER_TEMP\v12.2" --enable_cuda_profiling --use_vcpkg --use_vcpkg_ms_internal_asset_cache --enable_transformers_tool_test --cmake_extra_defines CMAKE_CUDA_ARCHITECTURES=86  --cmake_extra_defines onnxruntime_ENABLE_CUDA_EP_INTERNAL_TESTS=ON
+          python.exe ${{ github.workspace }}\tools\ci_build\build.py --update --build --config RelWithDebInfo --build_dir build --skip_submodule_sync --build_csharp --parallel --use_binskim_compliant_compile_flags --cmake_generator "Visual Studio 17 2022" --build_shared_lib --build_wheel --build_java --use_cuda --cuda_home="$env:RUNNER_TEMP\v12.8" --enable_cuda_profiling --use_vcpkg --use_vcpkg_ms_internal_asset_cache --enable_transformers_tool_test --cmake_extra_defines CMAKE_CUDA_ARCHITECTURES=86  --cmake_extra_defines onnxruntime_ENABLE_CUDA_EP_INTERNAL_TESTS=ON
           if ($lastExitCode -ne 0) {
             exit $lastExitCode
           }
@@ -188,20 +188,20 @@ jobs:
         working-directory: ${{ github.workspace }}
         shell: cmd
 
-      - name: Download CUDA SDK v12.2
+      - name: Download CUDA SDK v12.8
         working-directory: ${{ runner.temp }}
         run: |
-          azcopy.exe cp --recursive "https://lotusscus.blob.core.windows.net/models/cuda_sdk/v12.2" .
+          azcopy.exe cp --recursive "https://lotusscus.blob.core.windows.net/models/cuda_sdk/v12.8" .
           dir
         shell: pwsh
 
       - name: Add CUDA to PATH
         shell: powershell
         run: |
           Write-Host "Adding CUDA to PATH"
-          Write-Host "CUDA Path: $env:RUNNER_TEMP\v12.2\bin"
-          Add-Content -Path $env:GITHUB_PATH -Value "$env:RUNNER_TEMP\v12.2\bin"
-          Add-Content -Path $env:GITHUB_PATH -Value "$env:RUNNER_TEMP\v12.2\extras\CUPTI\lib64"
+          Write-Host "CUDA Path: $env:RUNNER_TEMP\v12.8\bin"
+          Add-Content -Path $env:GITHUB_PATH -Value "$env:RUNNER_TEMP\v12.8\bin"
+          Add-Content -Path $env:GITHUB_PATH -Value "$env:RUNNER_TEMP\v12.8\extras\CUPTI\lib64"
 
       - name: Set OnnxRuntimeBuildDirectory
         shell: pwsh
@@ -227,7 +227,7 @@ jobs:
             exit $lastExitCode
           }
 
-          python.exe ${{ github.workspace }}\tools\ci_build\build.py --test --config RelWithDebInfo --build_dir build --skip_submodule_sync --build_csharp --parallel --use_binskim_compliant_compile_flags --cmake_generator "Visual Studio 17 2022" --build_shared_lib --build_wheel --build_java --use_cuda --cuda_home="$env:RUNNER_TEMP\v12.2" --enable_cuda_profiling --use_vcpkg --use_vcpkg_ms_internal_asset_cache --enable_transformers_tool_test --cmake_extra_defines CMAKE_CUDA_ARCHITECTURES=86  --cmake_extra_defines onnxruntime_ENABLE_CUDA_EP_INTERNAL_TESTS=ON
+          python.exe ${{ github.workspace }}\tools\ci_build\build.py --test --config RelWithDebInfo --build_dir build --skip_submodule_sync --build_csharp --parallel --use_binskim_compliant_compile_flags --cmake_generator "Visual Studio 17 2022" --build_shared_lib --build_wheel --build_java --use_cuda --cuda_home="$env:RUNNER_TEMP\v12.8" --enable_cuda_profiling --use_vcpkg --use_vcpkg_ms_internal_asset_cache --enable_transformers_tool_test --cmake_extra_defines CMAKE_CUDA_ARCHITECTURES=86  --cmake_extra_defines onnxruntime_ENABLE_CUDA_EP_INTERNAL_TESTS=ON
           if ($lastExitCode -ne 0) {
             exit $lastExitCode
           }

.github/workflows/windows_openvino.yml

@@ -18,7 +18,7 @@ concurrency:
 jobs:
   BUILD_OPENVINO_EP:
     name: Windows OpenVINO CI Pipeline
-    runs-on: ["self-hosted", "1ES.Pool=onnxruntime-github-vs2022-mms"]
+    runs-on: ["self-hosted", "1ES.Pool=onnxruntime-github-vs2022-latest"]
     timeout-minutes: 240
     env:
       AZCOPY_AUTO_LOGIN_TYPE: MSI

.github/workflows/windows_qnn_x64.yml

@@ -18,7 +18,7 @@ concurrency:
 jobs:
   build_test_qnn_ep:
     name: Windows x64 QNN CI Pipeline (${{ matrix.QnnLibKind }})
-    runs-on: ["self-hosted", "1ES.Pool=onnxruntime-github-vs2022-mms"]
+    runs-on: ["self-hosted", "1ES.Pool=onnxruntime-github-vs2022-latest"]
     timeout-minutes: 120
     strategy:
       matrix: 

.github/workflows/windows_tensorrt.yml

@@ -19,7 +19,7 @@ concurrency:
 jobs:
   build:
     name: Windows GPU TensorRT CI Pipeline
-    runs-on: ["self-hosted", "1ES.Pool=onnxruntime-github-vs2022-mms"]
+    runs-on: ["self-hosted", "1ES.Pool=onnxruntime-github-vs2022-latest"]
     steps:
       - uses: actions/checkout@v5
         with:
@@ -41,10 +41,10 @@ jobs:
         working-directory: ${{ github.workspace }}
         shell: cmd
 
-      - name: Download CUDA SDK v12.2
+      - name: Download CUDA SDK v12.8
         working-directory: ${{ runner.temp }}
         run: |
-          azcopy.exe cp --recursive "https://lotusscus.blob.core.windows.net/models/cuda_sdk/v12.2" .
+          azcopy.exe cp --recursive "https://lotusscus.blob.core.windows.net/models/cuda_sdk/v12.8" .
           dir
         shell: pwsh
 
@@ -56,9 +56,9 @@ jobs:
         shell: powershell
         run: |
           Write-Host "Adding CUDA to PATH"
-          Write-Host "CUDA Path: $env:RUNNER_TEMP\v12.2\bin"
-          Add-Content -Path $env:GITHUB_PATH -Value "$env:RUNNER_TEMP\v12.2\bin"
-          Add-Content -Path $env:GITHUB_PATH -Value "$env:RUNNER_TEMP\v12.2\extras\CUPTI\lib64"
+          Write-Host "CUDA Path: $env:RUNNER_TEMP\v12.8\bin"
+          Add-Content -Path $env:GITHUB_PATH -Value "$env:RUNNER_TEMP\v12.8\bin"
+          Add-Content -Path $env:GITHUB_PATH -Value "$env:RUNNER_TEMP\v12.8\extras\CUPTI\lib64"
           Add-Content -Path $env:GITHUB_PATH -Value "$env:RUNNER_TEMP\TensorRT-10.9.0.34.Windows10.x86_64.cuda-12.8\lib"
 
       - uses: actions/setup-node@v5
@@ -116,7 +116,7 @@ jobs:
             exit $lastExitCode
           }
           # Execute the build process
-          python ${{ github.workspace }}\tools\ci_build\build.py --config RelWithDebInfo --parallel --use_binskim_compliant_compile_flags             --build_dir build --skip_submodule_sync --build_shared_lib --build --update --cmake_generator "Visual Studio 17 2022" --build_wheel --enable_onnx_tests             --use_tensorrt --tensorrt_home="${{ runner.temp }}\TensorRT-10.9.0.34.Windows10.x86_64.cuda-12.8"             --cuda_home="${{ runner.temp }}\v12.2" --use_vcpkg --use_vcpkg_ms_internal_asset_cache --cmake_extra_defines CMAKE_CUDA_ARCHITECTURES=86
+          python ${{ github.workspace }}\tools\ci_build\build.py --config RelWithDebInfo --parallel --use_binskim_compliant_compile_flags             --build_dir build --skip_submodule_sync --build_shared_lib --build --update --cmake_generator "Visual Studio 17 2022" --build_wheel --enable_onnx_tests             --use_tensorrt --tensorrt_home="${{ runner.temp }}\TensorRT-10.9.0.34.Windows10.x86_64.cuda-12.8"             --cuda_home="${{ runner.temp }}\v12.8" --use_vcpkg --use_vcpkg_ms_internal_asset_cache --cmake_extra_defines CMAKE_CUDA_ARCHITECTURES=86
           if ($lastExitCode -ne 0) {
             exit $lastExitCode
           }
@@ -193,10 +193,10 @@ jobs:
         working-directory: ${{ github.workspace }}
         shell: cmd
 
-      - name: Download CUDA SDK v12.2
+      - name: Download CUDA SDK v12.8
         working-directory: ${{ runner.temp }}
         run: |
-          azcopy.exe cp --recursive "https://lotusscus.blob.core.windows.net/models/cuda_sdk/v12.2" .
+          azcopy.exe cp --recursive "https://lotusscus.blob.core.windows.net/models/cuda_sdk/v12.8" .
           dir
         shell: pwsh
 
@@ -208,9 +208,9 @@ jobs:
         shell: powershell
         run: |
           Write-Host "Adding CUDA to PATH"
-          Write-Host "CUDA Path: $env:RUNNER_TEMP\v12.2\bin"
-          Add-Content -Path $env:GITHUB_PATH -Value "$env:RUNNER_TEMP\v12.2\bin"
-          Add-Content -Path $env:GITHUB_PATH -Value "$env:RUNNER_TEMP\v12.2\extras\CUPTI\lib64"
+          Write-Host "CUDA Path: $env:RUNNER_TEMP\v12.8\bin"
+          Add-Content -Path $env:GITHUB_PATH -Value "$env:RUNNER_TEMP\v12.8\bin"
+          Add-Content -Path $env:GITHUB_PATH -Value "$env:RUNNER_TEMP\v12.8\extras\CUPTI\lib64"
           Add-Content -Path $env:GITHUB_PATH -Value "$env:RUNNER_TEMP\TensorRT-10.9.0.34.Windows10.x86_64.cuda-12.8\lib"
 
       - name: Set OnnxRuntimeBuildDirectory
@@ -237,7 +237,7 @@ jobs:
             exit $lastExitCode
           }
 
-          python ${{ github.workspace }}\tools\ci_build\build.py --config RelWithDebInfo --parallel --use_binskim_compliant_compile_flags             --build_dir build --skip_submodule_sync --build_shared_lib --test --cmake_generator "Visual Studio 17 2022" --build_wheel --enable_onnx_tests             --use_tensorrt --tensorrt_home="${{ runner.temp }}\TensorRT-10.9.0.34.Windows10.x86_64.cuda-12.8"             --cuda_home="${{ runner.temp }}\v12.2" --use_vcpkg --use_vcpkg_ms_internal_asset_cache --cmake_extra_defines CMAKE_CUDA_ARCHITECTURES=86
+          python ${{ github.workspace }}\tools\ci_build\build.py --config RelWithDebInfo --parallel --use_binskim_compliant_compile_flags             --build_dir build --skip_submodule_sync --build_shared_lib --test --cmake_generator "Visual Studio 17 2022" --build_wheel --enable_onnx_tests             --use_tensorrt --tensorrt_home="${{ runner.temp }}\TensorRT-10.9.0.34.Windows10.x86_64.cuda-12.8"             --cuda_home="${{ runner.temp }}\v12.8" --use_vcpkg --use_vcpkg_ms_internal_asset_cache --cmake_extra_defines CMAKE_CUDA_ARCHITECTURES=86
           if ($lastExitCode -ne 0) {
             exit $lastExitCode
           }

.github/workflows/windows_x64_debug_build_x64_debug.yml

@@ -13,7 +13,7 @@ concurrency:
 
 jobs:
   build_x64_debug:
-    runs-on: ["self-hosted", "1ES.Pool=onnxruntime-github-vs2022-mms"]
+    runs-on: ["self-hosted", "1ES.Pool=onnxruntime-github-vs2022-latest"]
     timeout-minutes: 300
 
     steps:

.github/workflows/windows_x64_release_build_x64_release.yml

@@ -13,7 +13,7 @@ concurrency:
 
 jobs:
   build_x64_release:
-    runs-on: ["self-hosted", "1ES.Pool=onnxruntime-github-vs2022-mms"]
+    runs-on: ["self-hosted", "1ES.Pool=onnxruntime-github-vs2022-latest"]
     timeout-minutes: 300
 
     steps:

.github/workflows/windows_x64_release_ep_generic_interface_build_x64_release_ep_generic_interface.yml

@@ -13,7 +13,7 @@ concurrency:
 
 jobs:
   build_x64_release_ep_generic_interface:
-    runs-on: ["self-hosted", "1ES.Pool=onnxruntime-github-vs2022-mms"]
+    runs-on: ["self-hosted", "1ES.Pool=onnxruntime-github-vs2022-latest"]
     timeout-minutes: 300
 
     steps:

.github/workflows/windows_x64_release_vitisai_build_x64_release.yml

@@ -13,7 +13,7 @@ concurrency:
 
 jobs:
   build_x64_release_vitisai:
-    runs-on: ["self-hosted", "1ES.Pool=onnxruntime-github-vs2022-mms"]
+    runs-on: ["self-hosted", "1ES.Pool=onnxruntime-github-vs2022-latest"]
     timeout-minutes: 300
 
     steps:

.github/workflows/windows_x64_release_xnnpack.yml

@@ -13,7 +13,7 @@ concurrency:
 
 jobs:
   build_x64_release_xnnpack:
-    runs-on: ["self-hosted", "1ES.Pool=onnxruntime-github-vs2022-mms"]
+    runs-on: ["self-hosted", "1ES.Pool=onnxruntime-github-vs2022-latest"]
     timeout-minutes: 300
 
     steps:

.github/workflows/windows_x86.yml

@@ -13,7 +13,7 @@ concurrency:
 
 jobs:
   build_x86_release:
-    runs-on: ["self-hosted", "1ES.Pool=onnxruntime-github-vs2022-mms"]
+    runs-on: ["self-hosted", "1ES.Pool=onnxruntime-github-vs2022-latest"]
     timeout-minutes: 300
 
     steps:

VERSION_NUMBER

@@ -1 +1 @@
-1.23.0
+1.24.0

cmake/CMakeLists.txt

@@ -101,7 +101,7 @@ option(onnxruntime_USE_VSINPU "Build with VSINPU support" OFF)
 cmake_dependent_option(onnxruntime_USE_FLASH_ATTENTION "Build flash attention kernel for scaled dot product attention" ON "onnxruntime_USE_CUDA" OFF)
 option(onnxruntime_USE_LEAN_ATTENTION "Build lean attention kernel for scaled dot product attention" OFF)
 cmake_dependent_option(onnxruntime_USE_MEMORY_EFFICIENT_ATTENTION "Build memory efficient attention kernel for scaled dot product attention" ON "onnxruntime_USE_CUDA" OFF)
-cmake_dependent_option(onnxruntime_USE_FPA_INTB_GEMM "Build FpA IntB gemm cuda kernels" ON "onnxruntime_USE_CUDA" OFF)
+option(onnxruntime_USE_FPA_INTB_GEMM "Build FpA IntB gemm cuda kernels" OFF)
 
 option(onnxruntime_BUILD_FOR_NATIVE_MACHINE "Enable this option for turning on optimization specific to this machine" OFF)
 option(onnxruntime_USE_AVX "Use AVX instructions" OFF)
@@ -287,9 +287,13 @@ if (onnxruntime_ENABLE_TRAINING_APIS)
 endif()
 
 
-# Single output director for all binaries
+# Single output directory for all binaries
 set(RUNTIME_OUTPUT_DIRECTORY ${CMAKE_BINARY_DIR}/bin CACHE PATH "Single output directory for all binaries.")
 
+# Local mirror directory of cmake dependencies
+set(REPO_ROOT ${PROJECT_SOURCE_DIR}/..)
+set(onnxruntime_CMAKE_DEPS_MIRROR_DIR ${REPO_ROOT}/mirror CACHE PATH "Path to the local mirror of cmake dependencies")
+
 
 include(FetchContent)
 
@@ -425,7 +429,6 @@ if (onnxruntime_EXTENDED_MINIMAL_BUILD AND NOT onnxruntime_MINIMAL_BUILD)
   set(onnxruntime_MINIMAL_BUILD ON)
 endif()
 
-set(REPO_ROOT ${PROJECT_SOURCE_DIR}/..)
 set(ONNXRUNTIME_ROOT ${PROJECT_SOURCE_DIR}/../onnxruntime)
 set(ORTTRAINING_ROOT ${PROJECT_SOURCE_DIR}/../orttraining)
 set(ORTTRAINING_SOURCE_DIR ${ORTTRAINING_ROOT}/orttraining)

cmake/deps.txt

@@ -34,7 +34,7 @@ microsoft_gsl;https://github.com/microsoft/GSL/archive/refs/tags/v4.0.0.zip;cf36
 microsoft_wil;https://github.com/microsoft/wil/archive/refs/tags/v1.0.230629.1.zip;e4a542a323c070376f7c2d1973d0f7ddbc1d2fa5
 mimalloc;https://github.com/microsoft/mimalloc/archive/refs/tags/v2.1.1.zip;d5ee7d34223d0567892db5179849939c8769dc41
 mp11;https://github.com/boostorg/mp11/archive/refs/tags/boost-1.82.0.zip;9bc9e01dffb64d9e0773b2e44d2f22c51aace063
-onnx;https://github.com/onnx/onnx/archive/refs/tags/v1.19.0.zip;4c798b73e131438c196e6dcb9f3393968a8936f1
+onnx;https://github.com/onnx/onnx/archive/refs/tags/v1.19.1.zip;c5215b5697dcdfd71799f001b8c4054a6bba6b09
 # Use the latest commit of 10.9-GA
 onnx_tensorrt;https://github.com/onnx/onnx-tensorrt/archive/d5dce67db7c2e64b07e055571f5ec06f7f254de2.zip;01114d3b67650857281fa50faa2e412130a63b69
 protobuf;https://github.com/protocolbuffers/protobuf/archive/refs/tags/v21.12.zip;7cf2733949036c7d52fda017badcab093fe73bfa

cmake/external/helper_functions.cmake

@@ -4,11 +4,11 @@
 # 2. Set the cmake property COMPILE_WARNING_AS_ERROR to OFF for these external projects.
 
 function(onnxruntime_fetchcontent_declare contentName)
+    cmake_parse_arguments(PARSE_ARGV 1 ARG "" "URL;SOURCE_SUBDIR" "")
+    message(STATUS "Fetch ${contentName} from ${ARG_URL}")
     FetchContent_Declare(${ARGV})
     string(TOLOWER ${contentName} contentNameLower)
-    list(FIND ARGN SOURCE_SUBDIR index_SOURCE_SUBDIR)
-    if(index_SOURCE_SUBDIR GREATER_EQUAL 0)
-      cmake_parse_arguments(PARSE_ARGV 1 ARG "" "SOURCE_SUBDIR" "") 
+    if(NOT "${ARG_SOURCE_SUBDIR}" STREQUAL "")
       set(onnxruntime_${contentNameLower}_cmake_src_dir "${ARG_SOURCE_SUBDIR}" PARENT_SCOPE)
     endif()
 endfunction()

cmake/external/onnxruntime_external_deps.cmake

@@ -20,7 +20,7 @@ foreach(ONNXRUNTIME_DEP IN LISTS ONNXRUNTIME_DEPS_LIST)
 
     if(ONNXRUNTIME_DEP_URL MATCHES "^https://")
       # Search a local mirror folder
-      string(REGEX REPLACE "^https://" "${REPO_ROOT}/mirror/" LOCAL_URL "${ONNXRUNTIME_DEP_URL}")
+      string(REGEX REPLACE "^https://" "${onnxruntime_CMAKE_DEPS_MIRROR_DIR}/" LOCAL_URL "${ONNXRUNTIME_DEP_URL}")
 
       if(EXISTS "${LOCAL_URL}")
         cmake_path(ABSOLUTE_PATH LOCAL_URL)
@@ -498,13 +498,7 @@ else()
 endif()
 
 if(Patch_FOUND)
-  set(ONNXRUNTIME_ONNX_PATCH_COMMAND
-      ${Patch_EXECUTABLE} --binary --ignore-whitespace -p1 < ${PROJECT_SOURCE_DIR}/patches/onnx/onnx.patch &&
-      # Patch changes from https://github.com/onnx/onnx/pull/7253 to avoid unnecessary rebuilding.
-      # This change should be included in ONNX 1.19.1.
-      ${Patch_EXECUTABLE} --binary --ignore-whitespace -p1 <
-          ${PROJECT_SOURCE_DIR}/patches/onnx/avoid_regenerating_proto_files.patch
-      )
+  set(ONNXRUNTIME_ONNX_PATCH_COMMAND ${Patch_EXECUTABLE} --binary --ignore-whitespace -p1 < ${PROJECT_SOURCE_DIR}/patches/onnx/onnx.patch)
 else()
   set(ONNXRUNTIME_ONNX_PATCH_COMMAND "")
 endif()

cmake/onnxruntime_providers_cuda.cmake

@@ -182,8 +182,8 @@
 
     # Since CUDA 12.8, compiling diagnostics become stricter
     if (CMAKE_CUDA_COMPILER_VERSION VERSION_GREATER_EQUAL 12.8)
-      target_compile_options(${target} PRIVATE "$<$<COMPILE_LANGUAGE:CUDA>:--relocatable-device-code=true>")
-      set_target_properties(${target} PROPERTIES CUDA_SEPARABLE_COMPILATION ON)
+      target_compile_options(${target} PRIVATE "$<$<COMPILE_LANGUAGE:CUDA>:--static-global-template-stub=false>")
+
       if (MSVC)
         target_compile_options(${target} PRIVATE "$<$<COMPILE_LANGUAGE:CUDA>:SHELL:-Xcompiler /wd4505>")
       endif()

cmake/onnxruntime_test_pch.cmake

@@ -5,9 +5,11 @@ if(CMAKE_CXX_COMPILER_ID MATCHES "MSVC")
   target_precompile_headers(onnxruntime_test_all PRIVATE
     "${CMAKE_CURRENT_SOURCE_DIR}/test_pch.h"
   )
-  target_precompile_headers(onnxruntime_provider_test PRIVATE
-    "${CMAKE_CURRENT_SOURCE_DIR}/test_pch.h"
-  )
+  if (TARGET onnxruntime_provider_test)
+    target_precompile_headers(onnxruntime_provider_test PRIVATE
+      "${CMAKE_CURRENT_SOURCE_DIR}/test_pch.h"
+    )
+  endif()
 endif()
 
 # Exclude certain files that might conflict with PCH

cmake/onnxruntime_unittests.cmake

@@ -1228,6 +1228,11 @@ block()
     LIBS ${onnxruntime_provider_test_libs}
     DEPENDS ${onnxruntime_provider_test_deps}
   )
+  if (UNIX AND (onnxruntime_USE_TENSORRT OR onnxruntime_USE_NV))
+    # The test_main.cc includes NvInfer.h where it has many deprecated declarations
+    # simply ignore them for TensorRT EP build
+    set_property(TARGET onnxruntime_provider_test APPEND_STRING PROPERTY COMPILE_FLAGS "-Wno-deprecated-declarations")
+  endif()
 
   # enable dynamic plugin EP usage
   target_compile_definitions(onnxruntime_provider_test PRIVATE ORT_UNIT_TEST_ENABLE_DYNAMIC_PLUGIN_EP_USAGE)