Transition to Hermetic CUDA #8665
Draft
+66
−30
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
ysiraichi
force-pushed
the
ysiraichi/hermetic-cuda
branch
from
04cc3f2 to
d8ab058
Compare
|
Update: builds were successful, but tests are failing
|
Merged
ysiraichi
force-pushed
the
ysiraichi/hermetic-cuda
branch
2 times, most recently
from
4d3a6d0 to
ed49b54
Compare
This comment was marked as outdated.
This comment was marked as outdated.
ysiraichi
force-pushed
the
ysiraichi/hermetic-cuda
branch
from
0329124 to
952ed12
Compare
ysiraichi
force-pushed
the
ysiraichi/hermetic-cuda
branch
from
1b2b32e to
321bb36
Compare
|
can we add a description and value prop for this bug @ysiraichi? |
This was referenced Apr 29, 2025
- Update bazel files. - Use `clang` for PyTorch/XLA only. - Fix StableHLO tests. - Compile debugging information with dwarf-4.
ysiraichi
force-pushed
the
ysiraichi/hermetic-cuda
branch
from
bc64b8a to
ba8b262
Compare
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
This PR transitions PyTorch/XLA to use OpenXLA Hermetic CUDA for compiling PyTorch/XLA CUDA plugin. Such a change is necessary since the non-hermetic CUDA compilation was flagged as deprecated.
In order to do that, the following changes were necessary:
{Tpu,Gpu}CustomCallconstructors due to a use-after-movemlir::TypeID::get<[PassName]>(), e.g.BuildStableHLOCompositePassin stablehlo_composite_helper.cc