Merge branch 'main' into yifanl/trt_state

.github/workflows/rust-ci.yml

@@ -24,7 +24,7 @@ jobs:
     name: Download prebuilt ONNX Runtime archive from build.rs
     runs-on: ubuntu-latest
     env:
-      ORT_RUST_STRATEGY=download
+      ORT_RUST_STRATEGY: download
     steps:
       - uses: actions/checkout@v4
       - uses: ./.github/actions/rust-toolchain-setup

.github/workflows/stale.yml

@@ -29,7 +29,7 @@ jobs:
           # Label you want to apply to issues that have been inactive for the amount of time specified by days-before-issue-stale
           stale-issue-label: "stale"
           # Comment that you want to add to issues that are labeled by the actions/stale action
-          stale-issue-message: "This issue has been automatically marked as stale due to inactivity and will be closed in 7 days if no further activity occurs. If further support is needed, please provide an update and/or more details."
+          stale-issue-message: "This issue has been automatically marked as stale due to inactivity and will be closed in 30 days if no further activity occurs. If further support is needed, please provide an update and/or more details."
           # Comment that you want to add to issues that are closed by the actions/stale action
           close-issue-message: "This issue has been automatically closed due to inactivity. Please reactivate if further support is needed."
           # If you never want this action to label PRs, set this value to -1

.github/workflows/windows.yml

@@ -49,13 +49,10 @@ jobs:
       - uses: actions/checkout@v4
         with:
           submodules: true
-      - uses: actions/setup-python@v4
-        with:
-          python-version: '3.8.x'
-          architecture: 'x64'
       - uses: conda-incubator/setup-miniconda@v2
         with:
-          activate-environment: ""
+          activate-environment: "ort_build"
+          python-version: 3.8
       - name: 'Install LLVM-Dev'
         shell: pwsh
         run: |

.gitignore

@@ -195,3 +195,4 @@ Package.pins
 Package.resolved
 .build/
 .swiftpm/
+repros/

.pipelines/windowsai-steps.yml

@@ -84,7 +84,7 @@ jobs:
         7z x cmake-3.26.3-windows-x86_64.zip
         set PYTHONHOME=$(Build.BinariesDirectory)\${{ parameters.PythonPackageName }}.3.9.7\tools
         set PYTHONPATH=$(Build.BinariesDirectory)\${{ parameters.PythonPackageName }}.3.9.7\tools
-        $(Build.BinariesDirectory)\${{ parameters.PythonPackageName }}.3.9.7\tools\python.exe "$(Build.SourcesDirectory)\tools\ci_build\build.py" --build_dir $(Build.BinariesDirectory) --build_shared_lib --enable_onnx_tests --ms_experimental --use_dml --use_winml --cmake_generator "Visual Studio 17 2022" --update --config RelWithDebInfo --enable_lto --use_telemetry --disable_rtti --enable_wcos $(BuildFlags) --cmake_extra_defines CMAKE_SYSTEM_VERSION=10.0.19041.0 --cmake_path $(Build.BinariesDirectory)\cmake-3.26.3-windows-x86_64\bin\cmake.exe --ctest_path $(Build.BinariesDirectory)\cmake-3.26.3-windows-x86_64\bin\ctest.exe
+        $(Build.BinariesDirectory)\${{ parameters.PythonPackageName }}.3.9.7\tools\python.exe "$(Build.SourcesDirectory)\tools\ci_build\build.py" --build_dir $(Build.BinariesDirectory) --build_shared_lib --enable_onnx_tests --ms_experimental --use_dml --use_winml --cmake_generator "Visual Studio 17 2022" --update --config RelWithDebInfo --enable_lto --use_telemetry --disable_rtti --enable_wcos $(BuildFlags) --cmake_extra_defines "CMAKE_EXE_LINKER_FLAGS_RELWITHDEBINFO=/PROFILE" "CMAKE_SHARED_LINKER_FLAGS_RELWITHDEBINFO=/PROFILE" CMAKE_SYSTEM_VERSION=10.0.19041.0 --cmake_path $(Build.BinariesDirectory)\cmake-3.26.3-windows-x86_64\bin\cmake.exe --ctest_path $(Build.BinariesDirectory)\cmake-3.26.3-windows-x86_64\bin\ctest.exe
       workingDirectory: '$(Build.BinariesDirectory)'
       displayName: 'Generate cmake config'
 

README.md

@@ -30,7 +30,7 @@
 |Android|[![Build Status](https://dev.azure.com/onnxruntime/onnxruntime/_apis/build/status/Android%20CI%20Pipeline?label=Android)](https://dev.azure.com/onnxruntime/onnxruntime/_build/latest?definitionId=53)||
 |iOS|[![Build Status](https://dev.azure.com/onnxruntime/onnxruntime/_apis/build/status/iOS%20CI%20Pipeline?label=iOS)](https://dev.azure.com/onnxruntime/onnxruntime/_build/latest?definitionId=134)||
 |Web|[![Build Status](https://dev.azure.com/onnxruntime/onnxruntime/_apis/build/status/ONNX%20Runtime%20Web%20CI%20Pipeline?label=Web)](https://dev.azure.com/onnxruntime/onnxruntime/_build/latest?definitionId=161)||
-|Other|[![Build Status](https://dev.azure.com/onnxruntime/onnxruntime/_apis/build/status/onnxruntime-binary-size-checks-ci-pipeline?repoName=microsoft%2Fonnxruntime&label=Binary+Size+Check)](https://dev.azure.com/onnxruntime/onnxruntime/_build/latest?definitionId=187&repoName=microsoft%2Fonnxruntime)<br>[![Build Status](https://dev.azure.com/onnxruntime/onnxruntime/_apis/build/status/onnxruntime-python-checks-ci-pipeline?label=Python+Checks)](https://dev.azure.com/onnxruntime/onnxruntime/_build/latest?definitionId=164)||
+|Other|[![Build Status](https://dev.azure.com/onnxruntime/onnxruntime/_apis/build/status/onnxruntime-binary-size-checks-ci-pipeline?repoName=microsoft%2Fonnxruntime&label=Binary+Size+Check)](https://dev.azure.com/onnxruntime/onnxruntime/_build/latest?definitionId=187&repoName=microsoft%2Fonnxruntime)||
 
 ## Third-party Pipeline Status
 

build_arm64x.bat

@@ -0,0 +1,12 @@
+:: Copyright (c) Microsoft Corporation. All rights reserved.
+:: Licensed under the MIT License.
+
+@echo off
+
+setlocal
+set PATH=C:\Program Files\Git\usr\bin;%PATH%
+set LINK_REPRO_NAME=/mylink.rsp
+
+rem Requires a Python install to be available in your PATH
+python "%~dp0\tools\ci_build\build.py" --arm64 --buildasx  --build_dir "%~dp0\build\arm64-x" %*
+python "%~dp0\tools\ci_build\build.py" --arm64ec --buildasx --build_dir "%~dp0\build\arm64ec-x" %*

cgmanifests/generated/cgmanifest.json

@@ -36,7 +36,7 @@
       "component": {
         "type": "git",
         "git": {
-          "commitHash": "29bf8085f3bf17b84d30e34b3d7ff8248fda404e",
+          "commitHash": "dcd5bd5fd593e31465af3d9ef291d26c646b0a4f",
           "repositoryUrl": "https://github.com/abseil/abseil-cpp.git"
         },
         "comments": "abseil_cpp"
@@ -126,7 +126,7 @@
       "component": {
         "type": "git",
         "git": {
-          "commitHash": "f8d7d77c06936315286eb55f8de22cd23c188571",
+          "commitHash": "530d5c8c84abd2a46f38583ee817743c9b3a42b4",
           "repositoryUrl": "https://github.com/google/googletest.git"
         },
         "comments": "googletest"
@@ -316,7 +316,7 @@
       "component": {
         "type": "git",
         "git": {
-          "commitHash": "a4f72a314a85732ed67d5aa8d1088d207a7e0e61",
+          "commitHash": "5356c4a943a35e74d7cdc69486afcb8703b9a59a",
           "repositoryUrl": "https://github.com/ROCmSoftwarePlatform/composable_kernel.git"
         },
         "comments": "composable_kernel"

cmake/CMakeLists.txt

@@ -1258,13 +1258,7 @@ if (onnxruntime_USE_OPENVINO)
   endif()
 
   # Check OpenVINO version for support
-  if (${VER} MATCHES "2022.1" OR $ENV{INTEL_OPENVINO_DIR} MATCHES "2022.1")
-    set(OPENVINO_VERSION "2022.1")
-    add_definitions(-DOPENVINO_2022_1=1)
-  elseif (${VER} MATCHES "2022.2" OR $ENV{INTEL_OPENVINO_DIR} MATCHES "2022.2")
-    set(OPENVINO_VERSION "2022.2")
-    add_definitions(-DOPENVINO_2022_2=1)
-  elseif ($ENV{INTEL_OPENVINO_DIR} MATCHES "2022.3")
+  if ($ENV{INTEL_OPENVINO_DIR} MATCHES "2022.3")
     set(OPENVINO_VERSION "2022.3")
     add_definitions(-DOPENVINO_2022_3=1)
   elseif ($ENV{INTEL_OPENVINO_DIR} MATCHES "2023.0")
@@ -1273,9 +1267,12 @@ if (onnxruntime_USE_OPENVINO)
   elseif ($ENV{INTEL_OPENVINO_DIR} MATCHES "2023.1")
     set(OPENVINO_VERSION "2023.1")
     add_definitions(-DOPENVINO_2023_1=1)
-  elseif ($ENV{INTEL_OPENVINO_DIR} MATCHES "openvino")
-    set(OPENVINO_VERSION "2023.1")
+  elseif ($ENV{INTEL_OPENVINO_DIR} MATCHES "2023.2")
+    set(OPENVINO_VERSION "2023.2")
     add_definitions(-DOPENVINO_2023_1=1)
+  elseif ($ENV{INTEL_OPENVINO_DIR} MATCHES "openvino")
+    set(OPENVINO_VERSION "2023.2")
+    add_definitions(-DOPENVINO_2023_2=1)
   else()
     message(FATAL_ERROR "Unsupported OpenVINO version: ${INTEL_OPENVINO_DIR}")
   endif()
@@ -1587,6 +1584,13 @@ set(VERSION_STRING       "Internal Build" CACHE STRING "String representation of
 if (WIN32)
   list(APPEND onnxruntime_EXTERNAL_LIBRARIES ${SYS_PATH_LIB})
   list(APPEND onnxruntime_EXTERNAL_LIBRARIES debug Dbghelp)
+  # In a onecore build the umbrella libs already contains references to the APIs in advapi32, so in onecore build we do not need to link to advapi32
+  # In a non-onecore build, usually we also do not need to link to advapi32 because VC++ by default should have provide everything we need, except when the build target is Windows ARM32.
+  # In the future we will add a build option to allow users disabling all API uses from advapi32 because some Windows environments do not have these APIs. For example, some Windows do not have
+  # Windows Registry so we cannot query Registry values.
+  if(onnxruntime_target_platform STREQUAL "ARM" AND CMAKE_CXX_STANDARD_LIBRARIES MATCHES kernel32.lib)
+    list(APPEND onnxruntime_EXTERNAL_LIBRARIES advapi32)
+  endif()
 else()
   list(APPEND onnxruntime_EXTERNAL_LIBRARIES nsync::nsync_cpp)
   list(APPEND onnxruntime_EXTERNAL_LIBRARIES ${ICONV_LIB} ${CMAKE_DL_LIBS} Threads::Threads)
@@ -1776,3 +1780,8 @@ if(TARGET onnxruntime)
     "${PROJECT_BINARY_DIR}/${PROJECT_NAME}ConfigVersion.cmake"
     DESTINATION "${CMAKE_INSTALL_LIBDIR}/cmake/${PROJECT_NAME}")
 endif()
+
+if(DEFINED BUILD_AS_ARM64X)
+  set(ARM64X_TARGETS onnxruntime)
+  include("${CMAKE_SOURCE_DIR}/arm64x.cmake")
+endif()

cmake/arm64x.cmake

@@ -0,0 +1,33 @@
+set(arm64ReproDir "${CMAKE_SOURCE_DIR}/repros")
+
+if("${BUILD_AS_ARM64X}" STREQUAL "ARM64")
+	foreach (n ${ARM64X_TARGETS})
+		add_custom_target(mkdirs_${n} ALL COMMAND cmd /c (if exist \"${arm64ReproDir}/${n}_temp/\" rmdir /s /q \"${arm64ReproDir}/${n}_temp\") && mkdir \"${arm64ReproDir}/${n}_temp\" )
+		add_dependencies(${n} mkdirs_${n})
+		target_link_options(${n} PRIVATE "/LINKREPRO:${arm64ReproDir}/${n}_temp")
+		add_custom_target(${n}_checkRepro ALL COMMAND cmd /c if exist \"${n}_temp/*.obj\" if exist \"${n}\" rmdir /s /q \"${n}\" 2>nul && if not exist \"${n}\" ren \"${n}_temp\" \"${n}\" DEPENDS ${n}
+		WORKING_DIRECTORY ${arm64ReproDir})
+	endforeach()
+
+
+elseif("${BUILD_AS_ARM64X}" STREQUAL "ARM64EC")
+	foreach (n ${ARM64X_TARGETS})
+		set(ARM64_LIBS)
+		set(ARM64_OBJS)
+		set(ARM64_DEF)
+
+		file(GLOB ARM64_OBJS "${arm64ReproDir}/${n}/*.obj")
+		file(GLOB ARM64_DEF "${arm64ReproDir}/${n}/*.def")
+		file(GLOB ARM64_LIBS "${arm64ReproDir}/${n}/*.LIB")
+
+		if(NOT "${ARM64_DEF}" STREQUAL "")
+			set(ARM64_DEF "/defArm64Native:${ARM64_DEF}")
+		endif()
+		target_sources(${n} PRIVATE ${ARM64_OBJS})
+		target_link_options(${n} PRIVATE /machine:arm64x "${ARM64_DEF}")
+
+		if(NOT "${ARM64_LIBS}" STREQUAL "")
+			target_link_libraries(${n} PUBLIC ${ARM64_LIBS})
+		endif()
+	endforeach()
+endif()

cmake/deps.txt

@@ -12,7 +12,7 @@
 # NOTE: You must run deps_update_and_upload.py and generate_cgmanifest.py when ready to test your changes in a CI.
 # See https://microsoft.sharepoint.com/teams/ONNX2/_layouts/OneNote.aspx?id=%2Fteams%2FONNX2%2FShared%20Documents%2FNotebooks%2FONNX%20Ecosystem%20Team%20Notebook&wd=target%28Development.one%7C63D3AB47-51D1-4A62-9965-66882234BD44%2FAdd%20or%20update%20a%20dependency%20in%20deps.txt%7C0E9ED71D-89D5-40FA-B05F-C0123289C591%2F%29
 #
-abseil_cpp;https://github.com/abseil/abseil-cpp/archive/refs/tags/20230802.0.zip;04271dfbfac59269b6939e1e9d5faf0d18a7ba91
+abseil_cpp;https://github.com/abseil/abseil-cpp/archive/dcd5bd5fd593e31465af3d9ef291d26c646b0a4f.zip;6cc204586014e189f5c0fe3274f83162fa7c700c
 cxxopts;https://github.com/jarro2783/cxxopts/archive/3c73d91c0b04e2b59462f0a741be8c07024c1bc0.zip;6c6ca7f8480b26c8d00476e0e24b7184717fe4f0
 date;https://github.com/HowardHinnant/date/archive/refs/tags/v3.0.1.zip;2dac0c81dc54ebdd8f8d073a75c053b04b56e159
 dlpack;https://github.com/dmlc/dlpack/archive/refs/tags/v0.6.zip;4d565dd2e5b31321e5549591d78aa7f377173445
@@ -27,7 +27,7 @@ fp16;https://github.com/Maratyszcza/FP16/archive/0a92994d729ff76a58f692d3028ca1b
 fxdiv;https://github.com/Maratyszcza/FXdiv/archive/63058eff77e11aa15bf531df5dd34395ec3017c8.zip;a5658f4036402dbca7cebee32be57fb8149811e1
 google_benchmark;https://github.com/google/benchmark/archive/refs/tags/v1.7.0.zip;e97c368b176e8614e3f1bf13dd9abcf6a7ad9908
 google_nsync;https://github.com/google/nsync/archive/refs/tags/1.26.0.zip;5e7c00ef6bf5b787386fc040067903ec774e2752
-googletest;https://github.com/google/googletest/archive/refs/tags/v1.14.0.zip;0ac421f2ec11af38b0fff0f1992184032731a8bc
+googletest;https://github.com/google/googletest/archive/530d5c8c84abd2a46f38583ee817743c9b3a42b4.zip;5e3a61db2aa975cfd0f97ba92c818744e7fa7034
 googlexnnpack;https://github.com/google/XNNPACK/archive/0da379fc4808f9601faef392352018c741c0f297.zip;663883491e380b628e0a5b162b5f2658032fae73
 json;https://github.com/nlohmann/json/archive/refs/tags/v3.10.5.zip;f257f8dc27c5b8c085dc887b40cddd18ae1f725c
 microsoft_gsl;https://github.com/microsoft/GSL/archive/refs/tags/v4.0.0.zip;cf368104cd22a87b4dd0c80228919bb2df3e2a14

cmake/external/dnnl.cmake

@@ -25,6 +25,16 @@ elseif(onnxruntime_USE_DNNL AND onnxruntime_DNNL_GPU_RUNTIME STREQUAL "ocl" AND
   set(DNNL_GPU_CMAKE_ARGS "-DDNNL_GPU_RUNTIME=OCL " "-DOPENCLROOT=${onnxruntime_DNNL_OPENCL_ROOT}")
 endif()
 
+if(onnxruntime_USE_DNNL AND onnxruntime_DNNL_AARCH64_RUNTIME STREQUAL "acl" AND onnxruntime_DNNL_ACL_ROOT STREQUAL "")
+  message(FATAL_ERROR "--dnnl_acl_root required")
+elseif(onnxruntime_USE_DNNL AND onnxruntime_DNNL_AARCH64_RUNTIME STREQUAL "" AND NOT (onnxruntime_DNNL_ACL_ROOT STREQUAL ""))
+  message(FATAL_ERROR "--dnnl_aarch64_runtime required")
+elseif(onnxruntime_USE_DNNL AND onnxruntime_DNNL_AARCH64_RUNTIME STREQUAL "acl" AND NOT (onnxruntime_DNNL_ACL_ROOT STREQUAL ""))
+  file(TO_CMAKE_PATH ${onnxruntime_DNNL_ACL_ROOT} onnxruntime_DNNL_ACL_ROOT)
+  set(ACL_INCLUDE_DIR ${onnxruntime_DNNL_ACL_ROOT}/arm_compute)
+  set(DNNL_AARCH64_CMAKE_ARGS "-DDNNL_AARCH64_USE_ACL=ON")
+endif()
+
 if (onnxruntime_USE_DNNL)
   set(DNNL_SOURCE ${CMAKE_CURRENT_BINARY_DIR}/dnnl/src/dnnl/src)
   set(DNNL_INSTALL ${CMAKE_CURRENT_BINARY_DIR}/dnnl/install)
@@ -51,7 +61,7 @@ if (onnxruntime_USE_DNNL)
     GIT_TAG ${DNNL_TAG}
     # PATCH_COMMAND ${MKLDNN_PATCH_DISCARD_COMMAND} COMMAND ${DNNL_PATCH_COMMAND}
     SOURCE_DIR ${DNNL_SOURCE}
-    CMAKE_ARGS -DDNNL_BUILD_TESTS=OFF -DDNNL_ENABLE_CONCURRENT_EXEC=ON -DDNNL_BUILD_EXAMPLES=OFF -DCMAKE_BUILD_TYPE=${CMAKE_BUILD_TYPE} -DCMAKE_INSTALL_PREFIX=${DNNL_INSTALL} ${DNNL_GPU_CMAKE_ARGS}
+    CMAKE_ARGS -DDNNL_BUILD_TESTS=OFF -DDNNL_ENABLE_CONCURRENT_EXEC=ON -DDNNL_BUILD_EXAMPLES=OFF -DCMAKE_BUILD_TYPE=${CMAKE_BUILD_TYPE} -DCMAKE_INSTALL_PREFIX=${DNNL_INSTALL} ${DNNL_GPU_CMAKE_ARGS} ${DNNL_AARCH64_CMAKE_ARGS}
   )
   link_directories(${DNNL_LIB_DIR})
 endif()

cmake/external/onnxruntime_external_deps.cmake

@@ -47,8 +47,8 @@ if (onnxruntime_BUILD_UNIT_TESTS)
   FetchContent_Declare(
     googletest
     URL ${DEP_URL_googletest}
-    FIND_PACKAGE_ARGS 1.14.0...<2.0.0 NAMES GTest
     URL_HASH SHA1=${DEP_SHA1_googletest}
+    FIND_PACKAGE_ARGS 1.14.0...<2.0.0 NAMES GTest
   )
 endif()
 
@@ -124,7 +124,7 @@ if(CMAKE_CROSSCOMPILING AND NOT ONNX_CUSTOM_PROTOC_EXECUTABLE)
     if(protoc_binary_SOURCE_DIR)
       message("Use prebuilt protoc")
       set(ONNX_CUSTOM_PROTOC_EXECUTABLE ${protoc_binary_SOURCE_DIR}/bin/protoc.exe)
-	  set(PROTOC_EXECUTABLE ${ONNX_CUSTOM_PROTOC_EXECUTABLE})
+      set(PROTOC_EXECUTABLE ${ONNX_CUSTOM_PROTOC_EXECUTABLE})
     endif()
   elseif(CMAKE_HOST_SYSTEM_NAME STREQUAL "Linux")
     if(CMAKE_HOST_SYSTEM_PROCESSOR MATCHES "^(x86_64|amd64)$")
@@ -140,7 +140,7 @@ if(CMAKE_CROSSCOMPILING AND NOT ONNX_CUSTOM_PROTOC_EXECUTABLE)
     if(protoc_binary_SOURCE_DIR)
       message("Use prebuilt protoc")
       set(ONNX_CUSTOM_PROTOC_EXECUTABLE ${protoc_binary_SOURCE_DIR}/bin/protoc)
-	  set(PROTOC_EXECUTABLE ${ONNX_CUSTOM_PROTOC_EXECUTABLE})
+      set(PROTOC_EXECUTABLE ${ONNX_CUSTOM_PROTOC_EXECUTABLE})
     endif()
   elseif ((CMAKE_SYSTEM_NAME STREQUAL "Emscripten" OR CMAKE_SYSTEM_NAME STREQUAL "Android" OR CMAKE_SYSTEM_NAME STREQUAL "iOS") AND CMAKE_HOST_SYSTEM_NAME STREQUAL "Darwin")
     FetchContent_Declare(protoc_binary URL ${DEP_URL_protoc_mac_universal} URL_HASH SHA1=${DEP_SHA1_protoc_mac_universal})
@@ -281,7 +281,7 @@ if ((CPUINFO_SUPPORTED OR onnxruntime_USE_XNNPACK) AND NOT ANDROID)
     pytorch_clog
     URL ${DEP_URL_pytorch_cpuinfo}
     URL_HASH SHA1=${DEP_SHA1_pytorch_cpuinfo}
-	SOURCE_SUBDIR deps/clog
+    SOURCE_SUBDIR deps/clog
   )
   set(ONNXRUNTIME_CLOG_PROJ pytorch_clog)
   set(ONNXRUNTIME_CLOG_TARGET_NAME clog)

cmake/onnxruntime_mlas.cmake

@@ -284,6 +284,8 @@ else()
           set(X86 TRUE)
         elseif(CMAKE_SYSTEM_PROCESSOR MATCHES "^(x86_64|amd64)$")
           set(X86_64 TRUE)
+        elseif(CMAKE_SYSTEM_PROCESSOR MATCHES "^loongarch64.*")
+          set(LOONGARCH64 TRUE)
         endif()
     endif()
 
@@ -575,6 +577,26 @@ else()
           set(MLAS_SOURCE_IS_NOT_SET 0)
         endif()
     endif()
+    if(LOONGARCH64 AND MLAS_SOURCE_IS_NOT_SET)
+        set(mlas_platform_srcs
+          ${MLAS_SRC_DIR}/qgemm_kernel_lsx.cpp
+          ${MLAS_SRC_DIR}/loongarch64/SgemmKernelLasx.S
+          ${MLAS_SRC_DIR}/loongarch64/DgemmKernelLsx.S
+          ${MLAS_SRC_DIR}/loongarch64/DgemmKernelLasx.S
+          ${MLAS_SRC_DIR}/loongarch64/SgemmKernelLsx.S
+          ${MLAS_SRC_DIR}/loongarch64/SconvKernelLsx.S
+          ${MLAS_SRC_DIR}/loongarch64/SconvKernelLasx.S
+          ${MLAS_SRC_DIR}/loongarch64/SpoolKernelLSX.S
+          ${MLAS_SRC_DIR}/loongarch64/SpoolKernelLasx.S
+          ${MLAS_SRC_DIR}/loongarch64/SgemmTransposePackB16x4LSX.S
+          ${MLAS_SRC_DIR}/loongarch64/SgemmTransposePackB16x4Lasx.S
+          ${MLAS_SRC_DIR}/loongarch64/SoftmaxKernelLasx.S
+            )
+            set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -mlsx -mlasx")
+        if(NOT ONNXRUNTIME_MLAS_MULTI_ARCH)
+          set(MLAS_SOURCE_IS_NOT_SET 0)
+        endif()
+    endif()
     if(NOT ONNXRUNTIME_MLAS_MULTI_ARCH AND MLAS_SOURCE_IS_NOT_SET)
         file(GLOB_RECURSE mlas_platform_srcs
           "${MLAS_SRC_DIR}/scalar/*.cpp")

cmake/onnxruntime_providers_cuda.cmake

@@ -34,6 +34,8 @@
     if (NOT onnxruntime_USE_NCCL)
       list(REMOVE_ITEM onnxruntime_cuda_contrib_ops_cc_srcs
         "${ONNXRUNTIME_ROOT}/contrib_ops/cuda/collective/nccl_kernels.cc"
+        "${ONNXRUNTIME_ROOT}/contrib_ops/cuda/collective/sharded_moe.h"
+        "${ONNXRUNTIME_ROOT}/contrib_ops/cuda/collective/sharded_moe.cc"
         "${ONNXRUNTIME_ROOT}/contrib_ops/cuda/collective/sharding_spec.cc"
         "${ONNXRUNTIME_ROOT}/contrib_ops/cuda/collective/sharding.cc"
         "${ONNXRUNTIME_ROOT}/contrib_ops/cuda/collective/distributed_matmul.cc"

cmake/onnxruntime_providers_vitisai.cmake

@@ -15,16 +15,10 @@
     "${ONNXRUNTIME_ROOT}/core/providers/vitisai/imp/*.cc"
     "${ONNXRUNTIME_ROOT}/core/providers/vitisai/imp/*.h"
   )
-  list(REMOVE_ITEM onnxruntime_providers_vitisai_cc_srcs "${ONNXRUNTIME_ROOT}/core/providers/vitisai/onnxruntime_vitisai_ep_stub.cc")
   source_group(TREE ${ONNXRUNTIME_ROOT}/core FILES ${onnxruntime_providers_vitisai_cc_srcs})
   onnxruntime_add_static_library(onnxruntime_providers_vitisai ${onnxruntime_providers_vitisai_cc_srcs})
   onnxruntime_add_include_to_target(onnxruntime_providers_vitisai onnxruntime_common onnxruntime_framework onnx onnx_proto)
-  onnxruntime_add_shared_library(onnxruntime_vitisai_ep ${ONNXRUNTIME_ROOT}/core/providers/vitisai/onnxruntime_vitisai_ep_stub.cc)
-  onnxruntime_add_include_to_target(onnxruntime_vitisai_ep onnxruntime_common)
-  target_include_directories(onnxruntime_vitisai_ep PRIVATE "${ONNXRUNTIME_ROOT}" "${ONNXRUNTIME_ROOT}/core/providers/vitisai/include")
-  target_link_libraries(onnxruntime_providers_vitisai PUBLIC onnxruntime_vitisai_ep PRIVATE onnx protobuf::libprotobuf nlohmann_json::nlohmann_json )
-  target_compile_definitions(onnxruntime_vitisai_ep
-                           PRIVATE "-DONNXRUNTIME_VITISAI_EP_STUB=1" "-DONNXRUNTIME_VITISAI_EP_EXPORT_DLL=1")
+  target_link_libraries(onnxruntime_providers_vitisai PRIVATE onnx protobuf::libprotobuf nlohmann_json::nlohmann_json)
   if(NOT MSVC)
     target_compile_options(onnxruntime_providers_vitisai PUBLIC $<$<CONFIG:DEBUG>:-U_FORTIFY_SOURCE -D_FORTIFY_SOURCE=0>)
   endif(NOT MSVC)
@@ -49,4 +43,4 @@
             LIBRARY   DESTINATION ${CMAKE_INSTALL_LIBDIR}
             RUNTIME   DESTINATION ${CMAKE_INSTALL_BINDIR}
             FRAMEWORK DESTINATION ${CMAKE_INSTALL_BINDIR})
-  endif()
+  endif()