Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

Generic BLAS support on windows #351

Draft
wants to merge 45 commits into
base: main
Choose a base branch
from
Draft

Generic BLAS support on windows #351

wants to merge 45 commits into from

Conversation

h-vetinari
Copy link
Member

This had been de-scoped in #316, see 7269260

Targeting 2.6 (so based on #326), because that's going in as soon as CI resources allow.

@h-vetinari
pytorch v2.6.0.rc7
11db2ff
@h-vetinari
MNT: Re-rendered with conda-build 25.1.1, conda-smithy 3.45.2, and co…
5645301 
…nda-forge-pinning 2025.01.18.07.29.32
@h-vetinari
rebase patches
1219e26
@h-vetinari
Remove bound on setuptools version
1e211b2 
See pytorch/pytorch@2398e75

Partially reverts commit 46d7b46.
@h-vetinari
better sort pytorch run requirements
63c7c00
@h-vetinari
update dependencies for 2.6.0
bc3581d
@h-vetinari
pin libtorch in pytorch
113d413
@h-vetinari
backport fix for aarch build failure
c1b9cf6
@h-vetinari
reduce OMP_NUM_THREADS further due to OOMs in test
bf65aa4
@h-vetinari
use full paths to patched files in submodule
6757593
@h-vetinari
don't pin libtorch exactly for non-megabuilds
fa31c3f
@h-vetinari
make build number arithmetic explicit
49d134a 
don't rely on PKG_BUILDNUM resolving this correctly, which is either racy,
or implicitly depends on a separate render pass after setting build.number
@h-vetinari
disable patch in submodule
7cd981d
@h-vetinari
don't pin build number of libtorch in pytorch (for non-megabuild)
1405263 
otherwise conda breaks
```
conda_build.exceptions.RecipeError: Mismatching hashes in recipe. Exact pins in dependencies that contribute to the hash often cause this. Can you change one or more exact pins to version bound constraints?
Involved packages were:
Mismatching package: libtorch (id cpu_generic_habf3c96_0); dep: libtorch 2.6.0.rc7 *0; consumer package: pytorch
```
@h-vetinari
Merge remote-tracking branch 'upstream/main' into 2.6
51155b0
@h-vetinari
pytorch v2.6.0
be1ff5b
@h-vetinari
MNT: Re-rendered with conda-build 25.1.1, conda-smithy 3.45.4, and co…
95eb614 
…nda-forge-pinning 2025.01.28.22.50.30
@h-vetinari
remove test skips for 3.13
beb5481
@h-vetinari
vendor upstream smoke_test.py; not part of tarball
7c0d921
@h-vetinari
skip torchinductor tests on osx; segfaults hard
b627641 
and also runs into some openmp issue:
```
----------------------------- Captured stderr call -----------------------------
OMP: Error conda-forge#179: Function pthread_key_create failed:
OMP: System error conda-forge#35: Resource temporarily unavailable
```
@h-vetinari
backport patch for torchinductor failures on windows
c12f03d
@h-vetinari
skip a minor tolerance violation on osx
ce47e0a
@h-vetinari
drop unsuccessful patch; now covered by skip
ba2564b
@h-vetinari
drop unnecessary CUDAFLAGS in bld.bat
ab889b7 
this was done for consistency with `build.sh`, but appears unnecessary on windows
@h-vetinari
skip test_batchnorm_nhwc_cpu on linux as well
194f2d9
@h-vetinari
disable smoke test on aarch
8804810
@h-vetinari
Revert "Unvendor pybind11 and eigen"
93854a6 
keep it on windows though, where there's no issue

This reverts commit 9fcb3a7.
@h-vetinari
use different formulation for ptxas warning filter
b4ab27b
@h-vetinari
re-add zlib to pytorch's host deps
f72a13e
@h-vetinari
remove smoke test
b1658ef
@h-vetinari
deactivate pytest-xdist for linux-64+CUDA+MKL
8a43a63
@h-vetinari
temporary: skip known-passing builds to conserve resources
1da1037
@h-vetinari
Revert "deactivate pytest-xdist for linux-64+CUDA+MKL"
4784f25 
This reverts commit 8a43a63.
@h-vetinari
fix stray ) that broke pytest
51109f4
@h-vetinari
fix the way we look for $PREFIX/include in cross-compilation
796a52b
@h-vetinari
Revert "temporary: skip known-passing builds to conserve resources"
d72a0fd 
This reverts commit 1da1037.
@h-vetinari
Reapply "Unvendor pybind11 and eigen"
96d3954 
This reverts commit 93854a6.
@h-vetinari
different spelling for handing -w to -Xptxas
051b686
@h-vetinari
skip test_mutable_custom_op_fixed_layout everywhere
71e319d
@h-vetinari
robustify CONDA_BUILD_CROSS_COMPILATION handling
de7707c
@h-vetinari
set CUDA_VERSION during build; test torch.version.cuda is correct
381fcb8
@h-vetinari
enable torch.backends tests also for win+CUDA
04e1437
@h-vetinari
Revert "disable non-mkl builds on windows"
7549d5f 
This reverts commit 7269260.
@conda-forge-admin
Copy link
Contributor

conda-forge-admin commented Feb 9, 2025 •
edited
Loading

Hi! This is the friendly automated conda-forge-linting service.

I just wanted to let you know that I linted all conda-recipes in your PR (recipe/meta.yaml) and found it was in an excellent condition.

I do have some suggestions for making it better though...

For recipe/meta.yaml:

  • ℹ️ The recipe is not parsable by parser conda-souschef (grayskull). This parser is not currently used by conda-forge, but may be in the future. We are collecting information to see which recipes are compatible with grayskull.
  • ℹ️ The recipe is not parsable by parser conda-recipe-manager. The recipe can only be automatically migrated to the new v1 format if it is parseable by conda-recipe-manager.

This message was generated by GitHub Actions workflow run https://github.com/conda-forge/conda-forge-webservices/actions/runs/13230117408. Examine the logs at this URL for more detail.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Reviewers

@Tobias-Fischer Tobias-Fischer Awaiting requested review from Tobias-Fischer Tobias-Fischer will be requested when the pull request is marked ready for review Tobias-Fischer is a code owner

@baszalmstra baszalmstra Awaiting requested review from baszalmstra baszalmstra will be requested when the pull request is marked ready for review baszalmstra is a code owner

@beckermr beckermr Awaiting requested review from beckermr beckermr will be requested when the pull request is marked ready for review beckermr is a code owner

@benjaminrwilson benjaminrwilson Awaiting requested review from benjaminrwilson benjaminrwilson will be requested when the pull request is marked ready for review benjaminrwilson is a code owner

@hmaarrfk hmaarrfk Awaiting requested review from hmaarrfk hmaarrfk will be requested when the pull request is marked ready for review hmaarrfk is a code owner

@jeongseok-meta jeongseok-meta Awaiting requested review from jeongseok-meta jeongseok-meta will be requested when the pull request is marked ready for review jeongseok-meta is a code owner

@mgorny mgorny Awaiting requested review from mgorny mgorny will be requested when the pull request is marked ready for review mgorny is a code owner

@sodre sodre Awaiting requested review from sodre sodre will be requested when the pull request is marked ready for review sodre is a code owner

Assignees
No one assigned
Labels
None yet
Projects
None yet
Milestone
No milestone
Development

Successfully merging this pull request may close these issues.
2 participants
@h-vetinari @conda-forge-admin