diff options
| author | Chih-Hsuan Yen | 2021-12-31 14:06:49 +0800 |
|---|---|---|
| committer | Chih-Hsuan Yen | 2021-12-31 14:07:10 +0800 |
| commit | 0185531906bda3a9aba93bbb0f3dcfeb0ae671ad (patch) | |
| tree | c5101895f73be2e196ec65113b8dde92f72b256d /install-orttraining-files.diff | |
| parent | a84b5454e84aed96aa78cc311c407d496b3855d6 (diff) | |
| download | aur-0185531906bda3a9aba93bbb0f3dcfeb0ae671ad.tar.gz | |
various fixes/improvements
* Switch back from clang to gcc. Apparently upstream tests more on gcc
than on clang, and there are several compatibility issues between
onnxruntime and clang [1,2] as well as cuda and clang [3]. On the
other hand, internal compiler errors from gcc have been fixed.
* Add more optional dependencies for several sub-packages, as motivated
by [4].
* Fix missing orttraining Python files, which is discovered when I'm
checking optional dependencies.
* Don't hard-code usage of GNU make, as suggested in [4].
[1] https://github.com/microsoft/onnxruntime/pull/10014
[2] https://github.com/microsoft/onnxruntime/pull/10160
[3] https://forums.developer.nvidia.com/t/building-with-clang-cuda-11-3-0-works-but-with-cuda-11-3-1-fails-regression/182176
[4] https://aur.archlinux.org/packages/python-onnxruntime/#comment-843401
Diffstat (limited to 'install-orttraining-files.diff')
| -rw-r--r-- | install-orttraining-files.diff | 19 |
1 files changed, 19 insertions, 0 deletions
diff --git a/install-orttraining-files.diff b/install-orttraining-files.diff new file mode 100644 index 000000000000..e95601fcd183 --- /dev/null +++ b/install-orttraining-files.diff @@ -0,0 +1,19 @@ +--- a/setup.py 2021-12-29 22:44:09.924917943 +0800 ++++ b/setup.py 2021-12-29 22:49:16.216878004 +0800 +@@ -355,7 +355,7 @@ + 'Operating System :: Microsoft :: Windows', + 'Operating System :: MacOS']) + +-if enable_training: ++if True: + packages.extend(['onnxruntime.training', + 'onnxruntime.training.amp', + 'onnxruntime.training.optim', +@@ -373,6 +373,7 @@ + package_data['onnxruntime.training.ortmodule.torch_cpp_extensions.cuda.torch_gpu_allocator'] = ['*.cc'] + package_data['onnxruntime.training.ortmodule.torch_cpp_extensions.cuda.fused_ops'] = \ + ['*.cpp', '*.cu', '*.cuh', '*.h'] ++if enable_training: + requirements_file = "requirements-training.txt" + # with training, we want to follow this naming convention: + # stable: |