Conversation
Extend gpu arch microbenchmarking to emit mem_latency_us using a Triton dependent-load pointer chase over a 64 MiB index table. Adds --latency-only for fast smoke tests and wires the field into the arch JSON output. Co-authored-by: Cursor <cursoragent@cursor.com>
Avoid importing fp4fp6_helpers at module load; Triton dot_scaled JIT can abort on gfx90a when only mem_latency_us is needed. Co-authored-by: Cursor <cursoragent@cursor.com>
Codecov Report✅ All modified and coverable lines are covered by tests. 📢 Thoughts on this report? Let us know! |
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
2 participants
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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
TraceLens.PerfModel.benchmarking.microbenchusing a Triton dependent-load pointer chase (64 MiB index table).mem_latency_usin measured gpu arch JSON output; add--latency-onlyand--skip-latencyflags.fp4fp6_helpersso latency-only runs work on gfx90a/MI210 without MXFP4 Triton JIT abort.Validation
--latency-only --warmup 5 --rep 30).Test plan
pytest tests/test_perfmodel_benchmarking.py -k measured_archpython -m TraceLens.PerfModel.benchmarking.microbench --device 0 --latency-only --warmup 5 --rep 20mem_latency_usalongsidemem_bw_gbpsand MAF keysPairs with roofline PR that consumes
mem_latency_usforLATENCY_BOUNDclassification (separate PR).