mirror of
https://github.com/saymrwulf/onnxruntime.git
synced 2026-07-29 20:14:01 +00:00
Cherry-pick some PRs to the release branch (#15913)
### Description Cherry-pick the following PRs to the release branch https://github.com/microsoft/onnxruntime/pull/15858 https://github.com/microsoft/onnxruntime/pull/15807 https://github.com/microsoft/onnxruntime/pull/15816 https://github.com/microsoft/onnxruntime/pull/15775 https://github.com/microsoft/onnxruntime/pull/15779 They are all clean merges. --------- Co-authored-by: cloudhan <guangyunhan@microsoft.com> Co-authored-by: Sumit Agarwal <sumitagarwal330@gmail.com> Co-authored-by: RandySheriffH <48490400+RandySheriffH@users.noreply.github.com> Co-authored-by: Randy Shuai <rashuai@microsoft.com>
This commit is contained in:
Changming Sun
GitHub
parent
6ef420ec74
commit
2fd1aea33b
14 changed files with 40 additions and 16 deletions
|
|
@ -351,6 +351,31 @@ extends:
|
|||
- script: |
|
||||
dir $(Build.SourcesDirectory)\unzipped\runtimes\win-x64\_native
|
||||
|
||||
- task: EsrpCodeSigning@2
|
||||
displayName: "Sign Nuget package"
|
||||
inputs:
|
||||
ConnectedServiceName: 'OnnxRuntime CodeSign 20190817'
|
||||
FolderPath: $(Build.ArtifactStagingDirectory)
|
||||
Pattern: '*.nupkg'
|
||||
signConfigType: inlineSignParams
|
||||
inlineOperation: |
|
||||
[
|
||||
{
|
||||
"keyCode": "CP-401405",
|
||||
"operationSetCode": "NuGetSign",
|
||||
"parameters": [ ],
|
||||
"toolName": "sign",
|
||||
"toolVersion": "1.0"
|
||||
},
|
||||
{
|
||||
"keyCode": "CP-401405",
|
||||
"operationSetCode": "NuGetVerify",
|
||||
"parameters": [ ],
|
||||
"toolName": "sign",
|
||||
"toolVersion": "1.0"
|
||||
}
|
||||
]
|
||||
|
||||
- job: NuGet_Publishing
|
||||
pool:
|
||||
type: windows
|
||||
|
|
|
|||
|
|
@ -1,6 +1,6 @@
|
|||
<?xml version="1.0" encoding="utf-8"?>
|
||||
<packages>
|
||||
<package id="python" version="3.9.7" targetFramework="native" />
|
||||
<package id="Microsoft.AI.DirectML" version="1.10.1" targetFramework="native" />
|
||||
<package id="Microsoft.AI.DirectML" version="1.11.0" targetFramework="native" />
|
||||
<package id="Microsoft.Windows.CppWinRT" version="2.0.201201.7" targetFramework="native" />
|
||||
</packages>
|
||||
|
|
|
|||
|
|
@ -1,6 +1,6 @@
|
|||
<?xml version="1.0" encoding="utf-8"?>
|
||||
<packages>
|
||||
<package id="pythonx86" version="3.9.7" targetFramework="native" />
|
||||
<package id="Microsoft.AI.DirectML" version="1.10.1" targetFramework="native" />
|
||||
<package id="Microsoft.AI.DirectML" version="1.11.0" targetFramework="native" />
|
||||
<package id="Microsoft.Windows.CppWinRT" version="2.0.201201.7" targetFramework="native" />
|
||||
</packages>
|
||||
|
|
|
|||
|
|
@ -80,11 +80,11 @@ jobs:
|
|||
|
||||
# must call vsdevcmd first to add cmake to PATH
|
||||
- script: |
|
||||
curl -O -L https://github.com/Kitware/CMake/releases/download/v3.24.3/cmake-3.24.3-windows-x86_64.zip
|
||||
7z x cmake-3.24.3-windows-x86_64.zip
|
||||
curl -O -L https://github.com/Kitware/CMake/releases/download/v3.26.3/cmake-3.26.3-windows-x86_64.zip
|
||||
7z x cmake-3.26.3-windows-x86_64.zip
|
||||
set PYTHONHOME=$(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 16 2019" --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-3.24.3-windows-x86_64\bin\cmake.exe --ctest_path $(Build.BinariesDirectory)\cmake-3.24.3-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 16 2019" --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 $(Build.BinariesDirectory)\cmake-3.26.3-windows-x86_64\bin\ctest.exe
|
||||
workingDirectory: '$(Build.BinariesDirectory)'
|
||||
displayName: 'Generate cmake config'
|
||||
|
||||
|
|
|
|||
|
|
@ -272,6 +272,9 @@ if (onnxruntime_USE_ROCM)
|
|||
|
||||
file(GLOB rocm_cmake_components ${onnxruntime_ROCM_HOME}/lib/cmake/*)
|
||||
list(APPEND CMAKE_PREFIX_PATH ${rocm_cmake_components})
|
||||
# Force cmake to accept the configured HIP compiler. Because the configured CMAKE_PREFIX_PATH does not work during
|
||||
# enable_language(HIP), we might need to move configuring of CMAKE_PREFIX_PATH to build.py (in the future).
|
||||
set(CMAKE_HIP_COMPILER_FORCED ON)
|
||||
|
||||
enable_language(HIP)
|
||||
# NOTE: Flags -mllvm -amdgpu-early-inline-all=true are critical for gpu kernel code performance. -mllvm passes the
|
||||
|
|
|
|||
2
cmake/external/dml.cmake
vendored
2
cmake/external/dml.cmake
vendored
|
|
@ -41,7 +41,7 @@ if (NOT onnxruntime_USE_CUSTOM_DIRECTML)
|
|||
set(NUGET_CONFIG ${PROJECT_SOURCE_DIR}/../NuGet.config)
|
||||
set(PACKAGES_CONFIG ${PROJECT_SOURCE_DIR}/../packages.config)
|
||||
get_filename_component(PACKAGES_DIR ${CMAKE_CURRENT_BINARY_DIR}/../packages ABSOLUTE)
|
||||
set(DML_PACKAGE_DIR ${PACKAGES_DIR}/Microsoft.AI.DirectML.1.10.1)
|
||||
set(DML_PACKAGE_DIR ${PACKAGES_DIR}/Microsoft.AI.DirectML.1.11.0)
|
||||
|
||||
# Restore nuget packages, which will pull down the DirectML redist package.
|
||||
add_custom_command(
|
||||
|
|
|
|||
|
|
@ -1193,7 +1193,6 @@ Do not modify directly.*
|
|||
|DequantizeLinear|*in* x:**T1**<br> *in* x_scale:**T2**<br> *in* x_zero_point:**T1**<br> *out* y:**T2**|1+|**T1** = tensor(float)<br/> **T2** = tensor(uint8)|
|
||||
|EmbedLayerNormalization|*in* input_ids:**T1**<br> *in* segment_ids:**T1**<br> *in* word_embedding:**T**<br> *in* position_embedding:**T**<br> *in* segment_embedding:**T**<br> *in* gamma:**T**<br> *in* beta:**T**<br> *in* mask:**T1**<br> *in* position_ids:**T1**<br> *out* output:**T**<br> *out* mask_index:**T1**<br> *out* embedding_sum:**T**|1+|**T** = tensor(float), tensor(float16)|
|
||||
|FusedMatMul|*in* A:**T**<br> *in* B:**T**<br> *out* Y:**T**|1+|**T** = tensor(float), tensor(float16)|
|
||||
|FusedMatMulActivation|*in* A:**T**<br> *in* B:**T**<br> *out* Y:**T**|1+|**T** = tensor(float), tensor(float16)|
|
||||
|Gelu|*in* X:**T**<br> *out* Y:**T**|1+|**T** = tensor(float), tensor(float16)|
|
||||
|GroupNorm|*in* X:**T**<br> *in* gamma:**M**<br> *in* beta:**M**<br> *out* Y:**T**|1+|**M** = tensor(float), tensor(float16)<br/> **T** = tensor(float), tensor(float16)|
|
||||
|NhwcConv|*in* X:**T**<br> *in* W:**T**<br> *in* B:**T**<br> *out* Y:**T**|1+|**T** = tensor(float), tensor(float16)|
|
||||
|
|
|
|||
|
|
@ -182,7 +182,6 @@ InlinedVector<std::unique_ptr<GraphTransformer>> GenerateTransformers(
|
|||
#ifndef DISABLE_CONTRIB_OPS
|
||||
const InlinedHashSet<std::string_view> cpu_ep = {onnxruntime::kCpuExecutionProvider};
|
||||
#endif
|
||||
const InlinedHashSet<std::string_view> dml_ep = {onnxruntime::kDmlExecutionProvider};
|
||||
switch (level) {
|
||||
case TransformerLevel::Level1: {
|
||||
// RewriteRule optimizations are the simplest (they generally remove unnecessary nodes and are cheap to run)
|
||||
|
|
@ -309,7 +308,6 @@ InlinedVector<std::unique_ptr<GraphTransformer>> GenerateTransformers(
|
|||
transformers.emplace_back(std::make_unique<QuickGeluFusion>(cpu_cuda_dml_rocm_eps));
|
||||
|
||||
transformers.emplace_back(std::make_unique<MatMulScaleFusion>(cpu_cuda_dml_rocm_eps));
|
||||
transformers.emplace_back(std::make_unique<MatMulActivationFusion>(dml_ep));
|
||||
|
||||
// GeluApproximation has side effects which may change results. It needs to be manually enabled,
|
||||
// or alternatively the model can be updated offline using a model conversion script
|
||||
|
|
|
|||
|
|
@ -871,7 +871,6 @@ constexpr static OperatorRegistrationInformation operatorRegistrationInformation
|
|||
{REG_INFO_MS( 1, Gelu, typeNameListDefault, supportedTypeListFloat16to32, DmlGraphSupport::Supported)},
|
||||
{REG_INFO_MS( 1, BiasGelu, typeNameListDefault, supportedTypeListFloat16to32, DmlGraphSupport::Supported)},
|
||||
{REG_INFO_MS( 1, FusedMatMul, typeNameListDefault, supportedTypeListFloat16to32, DmlGraphSupport::Supported)},
|
||||
{REG_INFO_MS( 1, FusedMatMulActivation, typeNameListDefault, supportedTypeListFloat16to32, DmlGraphSupport::Supported)},
|
||||
{REG_INFO_MS( 1, QLinearSigmoid, typeNameListDefault, supportedTypeListQLinearSigmoid, DmlGraphSupport::Supported, requiredConstantCpuInputs(), std::nullopt, QueryQLinearSigmoid)},
|
||||
{REG_INFO_MS( 1, Attention, typeNameListAttention, supportedTypeListAttention, DmlGraphSupport::Supported, requiredConstantCpuInputs(), std::nullopt, QueryAttention)},
|
||||
|
||||
|
|
|
|||
|
|
@ -2827,7 +2827,7 @@ TEST(CApiTest, TestMultiStreamInferenceSimpleSSD) {
|
|||
}
|
||||
#endif
|
||||
|
||||
#if !defined(ORT_MINIMAL_BUILD)
|
||||
#if !defined(ORT_MINIMAL_BUILD) && !defined(REDUCED_OPS_BUILD)
|
||||
TEST(MultiKernelSingleSchemaTest, valid) {
|
||||
Ort::SessionOptions session_options;
|
||||
session_options.SetIntraOpNumThreads(1);
|
||||
|
|
|
|||
|
|
@ -1,6 +1,6 @@
|
|||
<?xml version="1.0" encoding="utf-8"?>
|
||||
<packages>
|
||||
<package id="Microsoft.AI.DirectML" version="1.10.1" targetFramework="native" />
|
||||
<package id="Microsoft.AI.DirectML" version="1.11.0" targetFramework="native" />
|
||||
<package id="Microsoft.Windows.CppWinRT" version="2.0.201201.7" targetFramework="native" />
|
||||
<package id="google.protobuf.tools" version="3.21.12" targetFramework="native" />
|
||||
</packages>
|
||||
|
|
|
|||
|
|
@ -24,9 +24,9 @@ RUN apt-get update && apt-get install --yes --no-install-recommends \
|
|||
unzip
|
||||
|
||||
# cmake
|
||||
RUN CMAKE_VERSION=3.25.2 && \
|
||||
RUN CMAKE_VERSION=3.26.3 && \
|
||||
aria2c -q -d /tmp -o cmake-${CMAKE_VERSION}-linux-x86_64.tar.gz \
|
||||
--checksum=sha-256=783da74f132fd1fea91b8236d267efa4df5b91c5eec1dea0a87f0cf233748d99 \
|
||||
--checksum=sha-256=28d4d1d0db94b47d8dfd4f7dec969a3c747304f4a28ddd6fd340f553f2384dc2 \
|
||||
https://github.com/Kitware/CMake/releases/download/v${CMAKE_VERSION}/cmake-${CMAKE_VERSION}-linux-x86_64.tar.gz && \
|
||||
tar -zxf /tmp/cmake-${CMAKE_VERSION}-linux-x86_64.tar.gz --strip=1 -C /usr
|
||||
|
||||
|
|
|
|||
|
|
@ -10,7 +10,7 @@ RUN cd /opt/mpi_install/ucx/build &&\
|
|||
make install
|
||||
|
||||
# CMake
|
||||
ENV CMAKE_VERSION=3.24.2
|
||||
ENV CMAKE_VERSION=3.26.3
|
||||
RUN cd /usr/local && \
|
||||
wget -q -O - https://github.com/Kitware/CMake/releases/download/v${CMAKE_VERSION}/cmake-${CMAKE_VERSION}-Linux-x86_64.tar.gz | tar zxf -
|
||||
ENV PATH=/usr/local/cmake-${CMAKE_VERSION}-linux-x86_64/bin:${PATH}
|
||||
|
|
|
|||
|
|
@ -190,7 +190,7 @@ def generate_repo_url(line_list, repo_url, commit_id):
|
|||
|
||||
|
||||
def generate_dependencies(xml_text, package_name, version):
|
||||
dml_dependency = '<dependency id="Microsoft.AI.DirectML" version="1.10.1"/>'
|
||||
dml_dependency = '<dependency id="Microsoft.AI.DirectML" version="1.11.0"/>'
|
||||
|
||||
if package_name == "Microsoft.AI.MachineLearning":
|
||||
xml_text.append("<dependencies>")
|
||||
|
|
|
|||
Loading…
Reference in a new issue