Gentoo Packages

Search

Install

Install this package:

emerge -a sci-libs/onnxruntime

Package Information

Description:: ONNX Runtime is Microsoft's cross-platform machine learning inference and training accelerator. Provides a C/C++/Python API plus optional execution providers for CPU, NVIDIA CUDA, AMD ROCm, OpenVINO, DirectML, CoreML, and others. Reads ONNX-format models and dispatches operators to whichever provider is registered. USE=python builds the Python bindings (NumPy, flatbuffers, protobuf, and packaging); USE=test builds the C++ test suite with upstream's pinned GoogleTest source. Derived from the ::guru package with patches to relax the flatbuffers dependency, drop -Werror, and use system libraries. Eigen3 remains pinned to a specific GitLab archive commit because upstream requires features newer than the latest compatible tagged release. CUDA builds use pinned CUTLASS and cuDNN Frontend sources required by upstream.
Homepage:: https://onnxruntime.ai https://github.com/microsoft/onnxruntime
License:: Apache-2.0 BSD MIT

Versions

Version	EAPI	Keywords	Slot
1.29.0-r2	8	~amd64 ~arm64	0
1.28.0-r2	8	~amd64 ~arm64	0

Metadata

Description

ONNX Runtime is Microsoft's cross-platform machine learning inference and training accelerator. Provides a C/C++/Python API plus optional execution providers for CPU, NVIDIA CUDA, AMD ROCm, OpenVINO, DirectML, CoreML, and others. Reads ONNX-format models and dispatches operators to whichever provider is registered. USE=python builds the Python bindings (NumPy, flatbuffers, protobuf, and packaging); USE=test builds the C++ test suite with upstream's pinned GoogleTest source. Derived from the ::guru package with patches to relax the flatbuffers dependency, drop -Werror, and use system libraries. Eigen3 remains pinned to a specific GitLab archive commit because upstream requires features newer than the latest compatible tagged release. CUDA builds use pinned CUTLASS and cuDNN Frontend sources required by upstream.

Maintainers

Ivan S. Titov (iohann.s.titov@gmail.com) - Type: person

Upstream

Bugs: https://github.com/microsoft/onnxruntime/issues
Remote IDs:
- github: microsoft/onnxruntime
- gitlab: libeigen/eigen

Raw Metadata XML

<pkgmetadata>
	<maintainer type="person">
		<email>iohann.s.titov@gmail.com</email>
		<name>Ivan S. Titov</name>
	</maintainer>
	<longdescription>
    ONNX Runtime is Microsoft's cross-platform machine learning inference
    and training accelerator. Provides a C/C++/Python API plus optional
    execution providers for CPU, NVIDIA CUDA, AMD ROCm, OpenVINO, DirectML,
    CoreML, and others. Reads ONNX-format models and dispatches operators
    to whichever provider is registered.

    USE=python builds the Python bindings (NumPy, flatbuffers, protobuf, and
    packaging); USE=test builds the C++ test suite with upstream's pinned
    GoogleTest source.

    Derived from the ::guru package with patches to relax the flatbuffers
    dependency, drop -Werror, and use system libraries. Eigen3 remains pinned
    to a specific GitLab archive commit because upstream requires features
    newer than the latest compatible tagged release. CUDA builds use pinned
    CUTLASS and cuDNN Frontend sources required by upstream.
  </longdescription>
	<use>
		<flag name="cuda">Build the NVIDIA CUDA execution provider</flag>
	</use>
	<upstream>
		<doc>https://onnxruntime.ai/docs/</doc>
		<bugs-to>https://github.com/microsoft/onnxruntime/issues</bugs-to>
		<remote-id type="github">microsoft/onnxruntime</remote-id>
		<remote-id type="gitlab">libeigen/eigen</remote-id>
	</upstream>
</pkgmetadata>

Lint Warnings

[Warning] USE flag 'python' in ebuild 1.29.0-r2 is not documented in metadata.xml
[Warning] USE flag 'test' in ebuild 1.29.0-r2 is not documented in metadata.xml
[Warning] USE flag 'python' in ebuild 1.28.0-r2 is not documented in metadata.xml
[Warning] USE flag 'test' in ebuild 1.28.0-r2 is not documented in metadata.xml
[Error] Manifest entry for unused DIST file 'abseil-cpp-20250814.1.tar.gz'
[Error] Manifest entry for unused DIST file 'cudnn-frontend-1.24.0.tar.gz'
[Error] Manifest entry for unused DIST file 'cutlass-4.4.2.tar.gz'
[Error] Manifest entry for unused DIST file 'eigen-3.4.0_p20250216.tar.bz2'
[Error] Manifest entry for unused DIST file 'googletest-1.17.0.tar.gz'
[Error] Manifest entry for unused DIST file 'onnxruntime-1.28.0.tar.gz'
[Error] Manifest entry for unused DIST file 'onnxruntime-1.29.0.tar.gz'
[Error] Repository is missing a repo-name in metadata/layout.conf or profiles/repo_name

USE Flags

Manage flags for this package: euse -i <flag> -p sci-libs/onnxruntime | euse -E <flag> -p sci-libs/onnxruntime | euse -D <flag> -p sci-libs/onnxruntime

Flag	Description	1.29.0-r2	1.28.0-r2
cuda	Build the NVIDIA CUDA execution provider	✓	✓
python	Build pygwy, the Python 3 scripting bridge. Implies USE=introspection and pulls in <pkg>dev-python/pygobject</pkg>. Provides a plug-in discovery path and an in-process console registered as /Python Console. ⚠️	✓	✓
test	Build and run unit tests ⚠️	✓	✓

Files

onnxruntime-1.22.2-relax-the-dependency-on-flatbuffers.patch
onnxruntime-1.24.4-no-werror.patch
onnxruntime-1.26.0-use-system-libraries.patch
onnxruntime-1.27.0-use-system-libraries.patch
onnxruntime-1.28.0-abseil-nvcc.patch
onnxruntime-1.28.0-use-system-libraries.patch
onnxruntime-1.29.0-fix-cuda-test-linking.patch

Manifest

Type	File	Size	Versions
DIST	abseil-cpp-20250814.1.tar.gz	2235716 bytes	1.29.0-r2, 1.28.0-r2
DIST	cudnn-frontend-1.24.0.tar.gz	6462724 bytes	1.29.0-r2, 1.28.0-r2
DIST	cutlass-4.4.2.tar.gz	39290487 bytes	1.29.0-r2, 1.28.0-r2
DIST	eigen-3.4.0_p20250216.tar.bz2	2272941 bytes	1.29.0-r2, 1.28.0-r2
DIST	googletest-1.17.0.tar.gz	885595 bytes	1.29.0-r2, 1.28.0-r2

Unmatched Entries

Type	File	Size
DIST	onnxruntime-1.28.0.tar.gz	278663126 bytes
DIST	onnxruntime-1.29.0.tar.gz	277662433 bytes

sci-libs/onnxruntime (stuff)