Update default branch post 25.01 (#296)

* Update CUDA archs in ORT (#291)

* Update CUDA archs in ORT

* Update CUDA archs in ORT and fixing typo

* Update cudnn home

* Version upgrade for openvino updated from 24.12

* Update CUDNN_home for v9.7.0

* Update the home path

* Update path

* Update CUDNN home

* Remove comments

* Update the supported igpu NVCC

* Update the supported igpu and dgpu NVCC

* Upadate default branch post 24.12 (#290)

* Update README and versions for 2.53.0 / 24.12 (#288)

* Update 'gen_ort_dockerfile.py' fil to meet changes in ONNX Runtime 1.20.x

* remove psutils

* Update reformating

* restore description

* Update OpenVINO to 2024.5 (#287)

* Update 'gen_ort_dockerfile.py' fil to meet changes in ONNX Runtime 1.20.x

* remove psutils

* Add OpenVINO version

* Fix pre-commit issues

* Extract archive in different location (#289)

* Extract archive in different location

* Revert "Extract archive in different location"

This reverts commit e57256a.

* Update installation instructions

* Reapply "Extract archive in different location"

This reverts commit 59e24e4.

* Rolling back changes cherry picked from 24.12

* Rolling back changes cherry picked from 24.12: fixing spaces

---------

Co-authored-by: Misha Chornyi <[email protected]>

* Downgrade patchelf version from 0.18.0 to 0.17.2  (#295)

Downgrade patchelf version from 0.18.0 to 0.17.2 due to patchelf regression

Patchelf shipped a regression in 0.18.0 and has since yanked the pypi release pointing to
0.17.2 as the most recent version. However, 0.18.0 is still the version shipped in both the apt and yum
repositories, thus we must use pip to install the version we want.
See mayeut/patchelf-pypi#87

---------

Co-authored-by: Pavithra Vijayakrishnan <[email protected]>
Co-authored-by: Kyle McGill <[email protected]>

Author: 3 people

cmake/download_onnxruntime.cmake

@@ -28,4 +28,5 @@ if(DEFINED TRITON_ONNXRUNTIME_PACKAGE_URL)
 
   endif(NOT DOWNLOAD_RESULT EQUAL 0)
 
-endif(DEFINED TRITON_ONNXRUNTIME_PACKAGE_URL)
+endif(DEFINED TRITON_ONNXRUNTIME_PACKAGE_URL)
+

tools/gen_ort_dockerfile.py

@@ -123,13 +123,13 @@ def dockerfile_for_linux(output_file):
         zip \
         ca-certificates \
         curl \
-        patchelf \
         python3-pip \
         git \
         gnupg \
         gnupg1 \
         openssl-devel
 
+RUN pip3 install patchelf==0.17.2
 """
     else:
         df += """
@@ -143,13 +143,14 @@ def dockerfile_for_linux(output_file):
         curl \
         libcurl4-openssl-dev \
         libssl-dev \
-        patchelf \
         python3-dev \
         python3-pip \
         git \
         gnupg \
         gnupg1
 
+RUN pip3 install patchelf==0.17.2
+
 # Install dependencies from
 # onnxruntime/dockerfiles/scripts/install_common_deps.sh.
 RUN apt update -q=2 \\
@@ -256,7 +257,7 @@ def dockerfile_for_linux(output_file):
         if FLAGS.cudnn_home is not None:
             ep_flags += ' --cudnn_home "{}"'.format(FLAGS.cudnn_home)
         elif target_platform() == "igpu":
-            ep_flags += ' --cudnn_home "/usr/lib/aarch64-linux-gnu"'
+            ep_flags += ' --cudnn_home "/usr/include"'
         if FLAGS.ort_tensorrt:
             ep_flags += " --use_tensorrt"
             if FLAGS.ort_version >= "1.12.1":
@@ -275,9 +276,9 @@ def dockerfile_for_linux(output_file):
         ep_flags += (
             " --skip_tests --cmake_extra_defines 'onnxruntime_BUILD_UNIT_TESTS=OFF'"
         )
-        cuda_archs = "53;62;72;87"
+        cuda_archs = "87;101"
     else:
-        cuda_archs = "75;80;86;90"
+        cuda_archs = "75;80;86;89;90;100;120"
 
     df += """
 WORKDIR /workspace/onnxruntime
@@ -472,7 +473,7 @@ def dockerfile_for_windows(output_file):
 WORKDIR /workspace/onnxruntime
 ARG VS_DEVCMD_BAT="\\BuildTools\\VC\\Auxiliary\\Build\\vcvars64.bat"
 RUN powershell Set-Content 'build.bat' -value 'call %VS_DEVCMD_BAT%',(Get-Content 'build.bat')
-RUN build.bat --cmake_generator "Visual Studio 17 2022" --config Release --cmake_extra_defines "CMAKE_CUDA_ARCHITECTURES=75;80;86;90" --skip_submodule_sync --parallel --build_shared_lib --compile_no_warning_as_error --skip_tests --update --build --build_dir /workspace/build {}
+RUN build.bat --cmake_generator "Visual Studio 17 2022" --config Release --cmake_extra_defines "CMAKE_CUDA_ARCHITECTURES=75;80;86;90;100;120" --skip_submodule_sync --parallel --build_shared_lib --compile_no_warning_as_error --skip_tests --update --build --build_dir /workspace/build {}
 """.format(
         ep_flags
     )