[ExecuTorch][WebGPU] Test coverage for the f16-accumulate steel GEMM by JCNTH · Pull Request #20801 · pytorch/executorch

JCNTH · 2026-07-09T03:44:55Z

Stack from ghstack (oldest at bottom):

Adds golden coverage for the f16-accumulate (pwdqf16acc) steel GEMM, which runs under WGPU_BACKEND_STEEL_F16ACC when group_size % BK == 0.

Key changes:

test_quantized_linear.py / test_webgpu_native.cpp — add pwdqf16acc (96x2048x256, K=2048) and pwdqf16acc_down (128x8192x2048, deep-K worst case) under #ifdef WGPU_BACKEND_STEEL_F16ACC, goldened against the fp64 dequant-matmul truth. f16 accumulation error grows with K, so the tolerances are wider than the f16-multiply steel_f16 (2.3e-4) and the deep-K shape gets the loosest gate; perplexity (the kernel diff) is the primary quality bar and this catches gross bit/index bugs.

Co-authored-with: Claude Code.
@exported-using-ghexport

Differential Revision: D111163651

[ghstack-poisoned]

pytorch-bot · 2026-07-09T03:44:58Z

🔗 Helpful Links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/executorch/20801

📄 Preview Python docs built from this PR

Note: Links to docs will display an error until the docs builds have been completed.

❌ 3 New Failures, 12 Pending, 2 Unrelated Failures

As of commit de9852c with merge base f4b01a8 ():

NEW FAILURES - The following jobs have failed:

Lint / lintrunner-mypy (gh)
>>> Lint for backends/cortex_m/quantizer/pattern_matcher.py:
pull / test-binary-size-linux-gcc / linux-job (gh)
RuntimeError: Command docker exec -t 13bf693715a7ccd6b438a224fd0a9974b0b7d1f0a44bc6cc5b61627df6d3465b /exec failed with exit code 1
pull / unittest / linux / linux-job (gh)
RuntimeError: Command docker exec -t df3e4001fecec4aa41163c7a556427f4e46a1601769910e422159e942406bf70 /exec failed with exit code 1

FLAKY - The following job failed but was likely due to flakiness present on trunk:

pull / unittest-editable / linux / linux-job (gh) (detected as infra flaky with no log or failing log classifier)

BROKEN TRUNK - The following job failed but was present on the merge base:

👉 Rebase onto the `viable/strict` branch to avoid these failures

pull / android / build-android (gh) (trunk failure)
ImportError: /opt/hostedtoolcache/Python/3.11.15/x64/lib/python3.11/site-packages/executorch/extension/training/pybindings/_training_lib.cpython-311-x86_64-linux-gnu.so: undefined symbol: _ZN3c104impl3cow23materialize_cow_storageERNS_11StorageImplE

This comment was automatically generated by Dr. CI and updates every 15 minutes.

github-actions · 2026-07-09T03:45:42Z

This PR needs a `release notes:` label

If your change should be included in the release notes (i.e. would users of this library care about this change?), please use a label starting with release notes:. This helps us keep track and include your important work in the next release notes.

To add a label, you can comment to pytorchbot, for example
@pytorchbot label "release notes: none"

For more information, see
https://github.com/pytorch/pytorch/wiki/PyTorch-AutoLabel-Bot#why-categorize-for-release-notes-and-how-does-it-work.

[ghstack-poisoned]

Pull Request resolved: #20801 Adds golden coverage for the f16-accumulate (`pwdqf16acc`) steel GEMM, which runs under `WGPU_BACKEND_STEEL_F16ACC` when `group_size % BK == 0`. Key changes: - `test_quantized_linear.py` / `test_webgpu_native.cpp` — add `pwdqf16acc` (96x2048x256, K=2048) and `pwdqf16acc_down` (128x8192x2048, deep-K worst case) under `#ifdef WGPU_BACKEND_STEEL_F16ACC`, goldened against the fp64 dequant-matmul truth. f16 accumulation error grows with K, so the tolerances are wider than the f16-multiply `steel_f16` (2.3e-4) and the deep-K shape gets the loosest gate; perplexity (the kernel diff) is the primary quality bar and this catches gross bit/index bugs. Co-authored-with: Claude Code. ghstack-source-id: 401485270 @exported-using-ghexport Differential Revision: [D111163651](https://our.internmc.facebook.com/intern/diff/D111163651/)

[ghstack-poisoned]

Pull Request resolved: #20801 Adds golden coverage for the f16-accumulate (`pwdqf16acc`) steel GEMM, which runs under `WGPU_BACKEND_STEEL_F16ACC` when `group_size % BK == 0`. Key changes: - `test_quantized_linear.py` / `test_webgpu_native.cpp` — add `pwdqf16acc` (96x2048x256, K=2048) and `pwdqf16acc_down` (128x8192x2048, deep-K worst case) under `#ifdef WGPU_BACKEND_STEEL_F16ACC`, goldened against the fp64 dequant-matmul truth. f16 accumulation error grows with K, so the tolerances are wider than the f16-multiply `steel_f16` (2.3e-4) and the deep-K shape gets the loosest gate; perplexity (the kernel diff) is the primary quality bar and this catches gross bit/index bugs. Co-authored-with: Claude Code. ghstack-source-id: 401485270 @exported-using-ghexport Differential Revision: [D111163651](https://our.internmc.facebook.com/intern/diff/D111163651/)

[ghstack-poisoned]

Pull Request resolved: #20801 Adds golden coverage for the f16-accumulate (`pwdqf16acc`) steel GEMM, which runs under `WGPU_BACKEND_STEEL_F16ACC` when `group_size % BK == 0`. Key changes: - `test_quantized_linear.py` / `test_webgpu_native.cpp` — add `pwdqf16acc` (96x2048x256, K=2048) and `pwdqf16acc_down` (128x8192x2048, deep-K worst case) under `#ifdef WGPU_BACKEND_STEEL_F16ACC`, goldened against the fp64 dequant-matmul truth. f16 accumulation error grows with K, so the tolerances are wider than the f16-multiply `steel_f16` (2.3e-4) and the deep-K shape gets the loosest gate; perplexity (the kernel diff) is the primary quality bar and this catches gross bit/index bugs. Co-authored-with: Claude Code. ghstack-source-id: 401515200 @exported-using-ghexport Differential Revision: [D111163651](https://our.internmc.facebook.com/intern/diff/D111163651/)

[ghstack-poisoned]

Pull Request resolved: #20801 Adds golden coverage for the f16-accumulate (`pwdqf16acc`) steel GEMM, which runs under `WGPU_BACKEND_STEEL_F16ACC` when `group_size % BK == 0`. Key changes: - `test_quantized_linear.py` / `test_webgpu_native.cpp` — add `pwdqf16acc` (96x2048x256, K=2048) and `pwdqf16acc_down` (128x8192x2048, deep-K worst case) under `#ifdef WGPU_BACKEND_STEEL_F16ACC`, goldened against the fp64 dequant-matmul truth. f16 accumulation error grows with K, so the tolerances are wider than the f16-multiply `steel_f16` (2.3e-4) and the deep-K shape gets the loosest gate; perplexity (the kernel diff) is the primary quality bar and this catches gross bit/index bugs. Co-authored-with: Claude Code. ghstack-source-id: 401515200 @exported-using-ghexport Differential Revision: [D111163651](https://our.internmc.facebook.com/intern/diff/D111163651/)

Update

06c3b12

[ghstack-poisoned]

JCNTH temporarily deployed to cadence July 9, 2026 03:44 — with GitHub Actions Inactive

meta-cla Bot added the CLA Signed This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed. label Jul 9, 2026

JCNTH temporarily deployed to cadence July 9, 2026 04:13 — with GitHub Actions Inactive

Update

1098506

[ghstack-poisoned]

JCNTH had a problem deploying to cadence July 9, 2026 17:41 — with GitHub Actions Error

Update

40108f1

[ghstack-poisoned]

JCNTH temporarily deployed to cadence July 9, 2026 17:53 — with GitHub Actions Inactive

meta-codesync Bot added the meta-exported label Jul 9, 2026

JCNTH had a problem deploying to cadence July 9, 2026 18:25 — with GitHub Actions Error

JCNTH had a problem deploying to cadence July 9, 2026 18:36 — with GitHub Actions Error

Update

b260184

[ghstack-poisoned]

JCNTH temporarily deployed to cadence July 9, 2026 18:59 — with GitHub Actions Inactive

JCNTH temporarily deployed to cadence July 9, 2026 19:27 — with GitHub Actions Inactive

Update

de9852c

[ghstack-poisoned]

JCNTH temporarily deployed to cadence July 9, 2026 20:30 — with GitHub Actions Inactive

JCNTH temporarily deployed to cadence July 9, 2026 20:58 — with GitHub Actions Inactive

meta-codesync Bot merged commit d47e216 into gh/JCNTH/16/base Jul 9, 2026
175 of 183 checks passed

meta-codesync Bot deleted the gh/JCNTH/16/head branch July 9, 2026 21:41

meta-codesync Bot temporarily deployed to cherry-pick-bot July 9, 2026 21:41 Inactive

pytorchbot mentioned this pull request Jul 9, 2026

[ExecuTorch][WebGPU] Test coverage for the f16-accumulate steel GEMM #20826

Merged

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

[ExecuTorch][WebGPU] Test coverage for the f16-accumulate steel GEMM#20801

[ExecuTorch][WebGPU] Test coverage for the f16-accumulate steel GEMM#20801
meta-codesync[bot] merged 6 commits into
gh/JCNTH/16/basefrom
gh/JCNTH/16/head

JCNTH commented Jul 9, 2026 •

edited

Loading

Uh oh!

pytorch-bot Bot commented Jul 9, 2026 •

edited

Loading

Uh oh!

github-actions Bot commented Jul 9, 2026

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

Uh oh!

Conversation

JCNTH commented Jul 9, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

pytorch-bot Bot commented Jul 9, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

🔗 Helpful Links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/executorch/20801

❌ 3 New Failures, 12 Pending, 2 Unrelated Failures

Uh oh!

github-actions Bot commented Jul 9, 2026

This PR needs a release notes: label

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

JCNTH commented Jul 9, 2026 •

edited

Loading

pytorch-bot Bot commented Jul 9, 2026 •

edited

Loading

This PR needs a `release notes:` label