Skip to content

Add dual-wave TBE benchmark and test harness - #6231

Open
q10 wants to merge 1 commit into
pytorch:mainfrom
q10:export-D116362668
Open

Add dual-wave TBE benchmark and test harness#6231
q10 wants to merge 1 commit into
pytorch:mainfrom
q10:export-D116362668

Conversation

@q10

@q10 q10 commented Aug 27, 2026

Copy link
Copy Markdown
Contributor

Summary:
Adds internal and OSS benchmark/test runners for the dual-wave TBE codegen work in D115263090, plus a canonical trace extractor and deterministic CPU coverage for the Jinja wave-set and dispatch helpers.

The benchmark harness records build and runtime wave metadata, supports same-artifact mixed gfx90a/gfx1100 validation, and emits canonical per-kernel statistics. The test harness covers existing TBE runtime targets, codegen type checks, and helper and emitted-dispatch assertions.

Differential Revision: D116362668

Summary:
Adds internal and OSS benchmark/test runners for the dual-wave TBE codegen work in D115263090, plus a canonical trace extractor and deterministic CPU coverage for the Jinja wave-set and dispatch helpers.

The benchmark harness records build and runtime wave metadata, supports same-artifact mixed gfx90a/gfx1100 validation, and emits canonical per-kernel statistics. The test harness covers existing TBE runtime targets, codegen type checks, and helper and emitted-dispatch assertions.

Differential Revision: D116362668
@meta-cla meta-cla Bot added the cla signed label Aug 27, 2026
@meta-codesync

meta-codesync Bot commented Aug 27, 2026

Copy link
Copy Markdown
Contributor

@q10 has exported this pull request. If you are a Meta employee, you can view the originating Diff in D116362668.

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

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant