[xpu][test][dtypes] Finish XPU enablement of test_affine_quantized_float - #4637
Open
karol-brejna-i wants to merge 1 commit into
Open
Conversation
Most tests in this file are already device-agnostic. Two gaps remain:
1. test_per_row_with_float32 was still CUDA-only:
- gate -> torch.accelerator.is_available()
- is_sm_at_least_89() scoped behind torch.cuda.is_available(), so the
sm89 requirement still applies to CUDA but does not exclude XPU
- .to("cuda") -> .to(get_current_accelerator_device()), resolved in
the test body
The assertion under test is raised in _input_activation_quant_func_fp8
after _fp8_mm_compat(), which only checks tensor rank and that both
dims are multiples of 16 - no device or vendor probe - so a 64x64
model raises it on any accelerator before any fp8 kernel runs.
2. Non-XPU fix: the gate on test_dequantize_affine_float8 compared a
bool to a string (torch.cuda.is_available() == "cuda"), so it was
always False and the sm89 requirement was never enforced on CUDA.
No effect on sm89+ or XPU; pre-sm89 NVIDIA now correctly skips.
test_quantize_dequantize_fp8_inductor is left alone: it is CPU-only by
construction and making it accelerator-aware needs XPU inductor codegen
expectations, which is separate scope.
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
🔗 Helpful Links🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/ao/4637
Note: Links to docs will display an error until the docs builds have been completed. ❌ You can merge normally! (5 Unrelated Failures), 1 Unclassified FailureAs of commit 72c043e with merge base 03ca489 ( UNCLASSIFIED FAILURE - DrCI could not classify the following job because the workflow did not run on the merge base. The failure may be pre-existing on trunk or introduced by this PR:
BROKEN TRUNK - The following jobs failed but were present on the merge base:👉 Rebase onto the `viable/strict` branch to avoid these failures
This comment was automatically generated by Dr. CI and updates every 15 minutes. |
meta-cla
Bot
added
the
CLA Signed
karol-brejna-i
marked this pull request as draft
July 29, 2026 07:58
liangan1
added
ciflow/xpu
Author
|
Need to wait until CI "global" fail will be over (see: #4625). |
karol-brejna-i
changed the title
karol-brejna-i
marked this pull request as ready for review
July 29, 2026 19:21
Contributor
|
affine_quantized_float is deprecated now I think, cc @andrewor14 @vkuzo maybe migrate useful tests to test_float8_tensor and remove the file? |
Author
|
@pytorchbot label "module: not user facing" |
pytorch-bot
Bot
added
the
module: not user facing
|
The ciflow label(s) ciflow/xpu will be added, but CI won't be triggered until the workflows are approved (scroll to the bottom of this page). Please ping one of the reviewers if you do not have access to approve and run workflows. |
liangan1
added
ciflow/xpu
|
|
liangan1
added
ciflow/xpu
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
3 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
Two targeted fixes to complete XPU enablement of
test/dtypes/test_affine_quantized_float.py:test_per_row_with_float32— last CUDA-hardcoded testReplaces the
torch.cuda.is_available()skip and hardcoded"cuda"device with the device-agnostic pattern.test_dequantize_affine_float8(line 144)The previous gate was:
torch.cuda.is_available()returnsbool, not a string, so this comparison is alwaysFalse. The sm89 requirement was never enforced on pre-sm89 CUDA hardware. This is a leftover from the earlier partial device-agnostic conversion (the sibling testtest_choose_scale_float8_boundsalready carries the correcttorch.cuda.is_available() and not is_sm_at_least_89()form). Fixed to the correct form.Validation
All checks performed on an XPU machine (torch 2.14.0.dev20260714+xpu):
ruff checkandruff format --checkboth clean. Grep sweep for.cuda()/device="cuda"/hardcoded skip strings returns nothing. Remainingtorch.cuda.is_available()occurrences are all properly paired withand not is_sm_at_least_89().