Package Details: python-onnxruntime 1.14.1-1

Git Clone URL: (read-only, click to copy)
Package Base: python-onnxruntime
Description: Cross-platform, high performance scoring engine for ML models
Upstream URL:
Licenses: MIT
Submitter: None
Maintainer: sl1pkn07
Last Packager: sl1pkn07
Votes: 5
Popularity: 0.64
First Submitted: 2019-07-12 11:06 (UTC)
Last Updated: 2023-05-23 06:53 (UTC)

Dependencies (43)

Sources (6)

Pinned Comments

yan12125 commented on 2021-01-26 08:35 (UTC) (edited on 2023-01-31 01:49 (UTC) by yan12125)

To build this package from sources, make sure you have enough free memory and disk space. On my machine (8-core Intel i7-4770), building this packages takes ~20GB disk space, ~2.3GB memory space and 40 minute build time. On a more powerful server (40-core Intel Xeon 4114), building takes ~18 minutes.

If you don't need CUDA, modify PKGBUILD and replace _ENABLE_CUDA=1 with _ENABLE_CUDA=0. Build time will be much shorter.

Latest Comments

« First ‹ Previous 1 2 3 4 5 6 7 Next › Last »

yan12125 commented on 2023-01-31 01:47 (UTC)

Thanks for the suggestion. As I've switched to PyPI wheels, I don't use this package anymore and I will orphan it. Feel free to adopt and improve it!

sl1pkn07 commented on 2023-01-30 19:09 (UTC)

something strange. the build of onnxruntime eats all my ram (128Gb). wtf XDDD

sl1pkn07 commented on 2023-01-30 18:27 (UTC)

please add instead off re2 in dependencies (or add the main package to makepepends and the library to depends). if not, pacman is unable to detect soversions bump when update the packages

└───╼  ldd /usr/lib/ (0x00007ffd56b33000) => /usr/lib/ (0x00007fbd05153000) => /usr/lib/ (0x00007fbd040d6000) => /usr/lib/ (0x00007fbd05146000) => not found


if the depends packages provides the name of librariy, please use them


yan12125 commented on 2022-10-28 16:23 (UTC)

I have easily build this and other packages with LTO enabled without any issues.

Thanks! I can confirm LTO works and I've updated PKGBUILD.

I just want to give you a feedback because I saw you disabled it in various packages.

I disabled LTO because it wasn't working. If it works now, feel free to leave a comment at corresponding packages.

Moebius14 commented on 2022-10-26 18:01 (UTC) (edited on 2022-10-26 19:28 (UTC) by Moebius14)

I get an error: invalid initialization of reference of type ‘const Eigen::half&’ from expression of type ‘int’ error in the file .../Eigen/src/Core/PartialReduxEvaluator.h when compiling. Nothing new, eigen always makes trouble when building onnx/tensorrt packages.

UPDATE:: Nevermind, I fixed the issue thanks to @bidskii who faced the exact same problem not so long ago. You need to patch the file mentioned above with this this commit, then everything will compile smoothly!

Thank you, bidskii!

Also @yan12125: I have easily build this and other packages with LTO enabled without any issues. I just want to give you a feedback because I saw you disabled it in various packages.

yan12125 commented on 2022-10-23 09:08 (UTC)


sl1pkn07 commented on 2022-10-22 23:59 (UTC)

please add -c protocol.file.allow=always in


yan12125 commented on 2022-08-27 04:34 (UTC)

@acxz: Thanks for noticing that! I've pushed a fix.

acxz commented on 2022-08-26 23:03 (UTC)

Building in a clean chroot, I am getting the following build error:

CMake Warning at CMakeLists.txt:1591 (find_package):
  By not providing "FindFlatbuffers.cmake" in CMAKE_MODULE_PATH this project
  has asked CMake to find a package configuration file provided by
  "Flatbuffers", but CMake did not find one.

  Could not find a package configuration file provided by "Flatbuffers" with
  any of the following names:


  Add the installation prefix of "Flatbuffers" to CMAKE_PREFIX_PATH or set
  "Flatbuffers_DIR" to a directory containing one of the above files.  If
  "Flatbuffers" provides a separate development package or SDK, be sure it
  has been installed.

Use flatbuffers from submodule
CMake Error at CMakeLists.txt:1607 (add_subdirectory):
  The source directory


  does not contain a CMakeLists.txt file.

yan12125 commented on 2022-03-30 12:44 (UTC)

Hmm, if you specify CPUExecutionProvider (or DnnlExecutionProvider if you have a recent Intel CPU) like, things should work without CUDA. Anyway, it indeed makes little sense to install giant cuda packages without using them. I will find time to add an option in PKGBUILD to disable CUDA builds. Your modifications seem a good start point :)