[TEST] Cover Gluon Blackwell tcgen05 ops - #459
Merged
Merged
Conversation
Jokeren
force-pushed
the
test/gluon-blackwell-tensor-memory-ops
branch
9 times, most recently
from
June 26, 2026 13:21
4d83822 to
60c2748
Compare
Jokeren
force-pushed
the
test/gluon-blackwell-tcgen05-ops
branch
from
June 26, 2026 14:48
6049438 to
b1422ef
Compare
Performance Benchmark
Iterations: 1 warmup + 20 measured |
Jokeren
marked this pull request as ready for review
June 26, 2026 16:14
mark14wu
added a commit
that referenced
this pull request
Jul 10, 2026
Resolves the PR #361 merge conflict (CI could not run on the merge ref). Main brings the merged #435 LoopSite refactor and the Gluon Blackwell test coverage (#456-#459). Resolutions: - race_detector.py _process_pending_check: keep the branch's _unsupported_capture early-return guard on top of main's version. - core/frontend/base.py: take main's simplified _LangPatchScope wholesale (3-tuple changes, no set_item/mark_removed) — the branch's 4-tuple machinery has no remaining callers on either side, and the auto-merge hybrid would have crashed restore(). - Adapt the race detector's loop hooks to #435's LoopSite keying: hook params and loop_stack matching use LoopSite (a hashable NamedTuple) instead of bare linenos; _finished_loop_iter_subs keys become LoopSites (two loops at the same function-relative line in different files no longer share bookkeeping). - simulation/gluon.py: broaden the optional gfx1250 import guard — a triton at the declared floor (>= 3.6.0) lacks the newer submodules (cluster/tdm/...) and the narrow is_hip_gfx1250-only guard killed every consumer at collection time on such installs. Full suite: 717 passed; the only delta vs the pre-merge baseline is inside the environment-dependent gluon family (4 old failures fixed by main, 3 new tests hit the local triton-3.6.0 x numpy-2 interpreter boundary that CI's environment does not).
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
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
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.
Summary
Add CPU-result coverage for Blackwell
tcgen05_mmasimulation through Gluon.This adds a focused end-to-end test file that:
blackwell.tcgen05_mmausing bothtcgen05_commitand explicit mbarrier signaling pathsa.float() @ b.float() + cValidation
PYTHONPATH=/mnt/keren/triton-viz pytest tests/end_to_end/test_gluon_blackwell_tcgen05_ops.py -qPYTHONPATH=/mnt/keren/triton-viz pytest tests/end_to_end/test_gluon_blackwell_tcgen05_ops.py tests/end_to_end/test_gluon_blackwell_tensor_memory_ops.py tests/end_to_end/test_gluon_wgmma_ops.py tests/end_to_end/test_gluon_tma_im2col_ops.py tests/end_to_end/test_gluon_blackwell_tma_ops.py tests/end_to_end/test_gluon_tma_ops.py tests/end_to_end/test_gluon_async_copy_ops.py tests/end_to_end/test_gluon_core_ops.py tests/end_to_end/test_gluon.py tests/end_to_end/test_core.py tests/unit/test_adapters.py -qpre-commit run --files tests/end_to_end/test_gluon_blackwell_tcgen05_ops.py