[cutlass-4][take 2] upgrade to cutlass 4.2.1#164159
[cutlass-4][take 2] upgrade to cutlass 4.2.1#164159henrylhtsang wants to merge 1 commit intopytorch:mainfrom
Conversation
🔗 Helpful Links🧪 See artifacts and rendered test results at hud.pytorch.org/pr/164159
Note: Links to docs will display an error until the docs builds have been completed. ❌ 2 New FailuresAs of commit aa16e5e with merge base 566ea4e ( NEW FAILURES - The following jobs have failed:
This comment was automatically generated by Dr. CI and updates every 15 minutes. |
|
@henrylhtsang has exported this pull request. If you are a Meta employee, you can view the originating diff in D83492704. |
| @@ -421,7 +421,7 @@ def get_max_alignment(inductor_layout: Layout) -> int: | |||
| offset = inductor_layout.offset | |||
|
|
|||
| def is_static_int(number): | |||
There was a problem hiding this comment.
| def is_static_int(number): | |
| def is_static_int(number: object) -> Type[int | sympy.Integer]: |
Otherwise this will hide the type information of the isinstance call...
| @@ -234,7 +234,6 @@ def _get_arg_from_node( | |||
| # node's memory, a stride tuple, the datatype | |||
| # Once again, need to check for local class type for stride tuple | |||
| if str(arg_ty) in { | |||
There was a problem hiding this comment.
This should be an equality check now
| @@ -173,7 +174,6 @@ def _render_argument_type( | |||
| # Fragile, but this is the only way to guarantee t is expected type because t is a local class | |||
| def is_nested_visitor_type(t: type) -> bool: | |||
| return ".".join([t.__module__, t.__qualname__]) in { | |||
| import cutlass_cppgen as python_cutlass # type: ignore[import-untyped, import-not-found] # noqa: F401 | ||
| else: | ||
| import cutlass_cppgen as python_cutlass # type: ignore[import-untyped, import-not-found] # noqa: F401 |
There was a problem hiding this comment.
| import cutlass_cppgen as python_cutlass # type: ignore[import-untyped, import-not-found] # noqa: F401 | |
| else: | |
| import cutlass_cppgen as python_cutlass # type: ignore[import-untyped, import-not-found] # noqa: F401 | |
| import cutlass_cppgen as python_cutlass # type: ignore[import-untyped, import-not-found] # noqa: F401 |
|
@Skylion007 unfortunately vllm side is very tricky. vllm-project/vllm#24673 |
Hmm, the other PR is currently merging though and missing some of these other import changes? So CUTLASS inductor will be broken after the other PR lands? Or are you saying VLLM will need to use an old version of CUTLASS temporarily. |
I think there are two upgrade going on:
|
Summary: Pull Request resolved: pytorch#164159 Test Plan: Sandcastle Differential Revision: D83492704
|
@henrylhtsang has exported this pull request. If you are a Meta employee, you can view the originating Diff in D83492704. |
e6be173 to
aa16e5e
Compare
|
@pytorchbot merge (Initiating merge automatically since Phabricator Diff has merged) |
Merge startedYour change will be merged once all checks pass (ETA 0-4 Hours). Learn more about merging in the wiki. Questions? Feedback? Please reach out to the PyTorch DevX Team |
Merge failedReason: 2 jobs have failed, first few of them are: inductor / unit-test / inductor-halide-build / build, inductor / unit-test / inductor-cpu-build / build Details for Dev Infra teamRaised by workflow job |
|
@pytorchbot merge -i |
Merge startedYour change will be merged while ignoring the following 2 checks: inductor / unit-test / inductor-halide-build / build, inductor / unit-test / inductor-cpu-build / build Learn more about merging in the wiki. Questions? Feedback? Please reach out to the PyTorch DevX Team |
…4589) Differential Revision: D83809105 Handle reviews from #164159 Pull Request resolved: #164589 Approved by: https://github.com/Skylion007
Test Plan: Sandcastle Differential Revision: D83492704 Pull Request resolved: pytorch#164159 Approved by: https://github.com/Skylion007, https://github.com/mlazos
…orch#164589) Differential Revision: D83809105 Handle reviews from pytorch#164159 Pull Request resolved: pytorch#164589 Approved by: https://github.com/Skylion007
Test Plan: Sandcastle
Differential Revision: D83492704
cc @voznesenskym @penguinwu @EikanWang @jgong5 @Guobing-Chen @XiaobingSuper @zhuhaozhe @blzheng @wenzhe-nrv @jiayisunx @ipiszy @chenyang78 @kadeng @muchulee8 @amjames @chauhang @aakhundov @coconutruben