Update third party tests - #3023
Open
antonwolfy wants to merge 18 commits into
Open
antonwolfy wants to merge 18 commits into
antonwolfy wants to merge 18 commits into
Conversation
Contributor
|
View rendered docs @ https://intelpython.github.io/dpnp/pull/3023/index.html |
antonwolfy
marked this pull request as ready for review
August 13, 2026 12:32
antonwolfy
requested review from
ndgrigorian and
vlad-perevezentsev
as code owners
August 13, 2026 12:32
antonwolfy
marked this pull request as draft
August 13, 2026 12:32
Contributor
|
Array API standard conformance tests for dpnp=0.21.0dev11=py314ha0e2e8e_27 ran successfully. |
Collaborator
|
No base build to compare — update-third-party-tests into master |
antonwolfy
force-pushed
the
update-third-party-tests
branch
2 times, most recently
from
August 19, 2026 21:08
cb8c139 to
5bd33aa
Compare
antonwolfy
force-pushed
the
update-third-party-tests
branch
4 times, most recently
from
September 13, 2026 12:26
6683145 to
636a808
Compare
antonwolfy
force-pushed
the
update-third-party-tests
branch
3 times, most recently
from
September 29, 2026 11:56
971de36 to
930a5b8
Compare
test_cum{sum,prod}_axis_batch_kernels hardcoded float64 inputs, which
dpnp downcasts to float32 on devices without fp64 support while NumPy
keeps float64 -- tripping the allclose dtype check and, for cumprod,
diverging by percent-level near float32 underflow along long axes.
Build both sides at cupy.default_float_type() so they compute at the
same precision, and relax cumprod to rtol=1e-5/atol=1e-6 to absorb
float32 scan rounding.
On non-fp64 devices test_cumprod_axis_batch_kernels runs in float32, where the cumulative product overflows to inf along long axes (dpnp and the NumPy reference alike, so the inf==inf comparison still holds). Add a message-scoped filterwarnings mark so the expected NumPy overflow RuntimeWarning is no longer reported by the infra-warnings tracker.
antonwolfy
force-pushed
the
update-third-party-tests
branch
from
September 30, 2026 09:23
930a5b8 to
d0df49c
Compare
antonwolfy
marked this pull request as ready for review
September 30, 2026 10:48
ndgrigorian
reviewed
Sep 30, 2026
| @@ -511,11 +599,11 @@ def test_compiler_flag(self): | |||
| x1, x2, y = self._helper(ker_times, cupy.float64) | |||
Collaborator
There was a problem hiding this comment.
Looks like this test passes float64 in as dtype, is it a concern on devices w/o fp64?
ndgrigorian
reviewed
Sep 30, 2026
| ) | ||
|
|
||
| @testing.for_dtypes(dtype_) | ||
| @testing.for_dtypes([numpy.float64, numpy.complex128]) |
Collaborator
There was a problem hiding this comment.
Seems the has_support_aspect check was removed here, is it safe?
ndgrigorian
reviewed
Sep 30, 2026
| # @pytest.mark.thread_unsafe(reason="allocation too large.") | ||
| def test_argmax_int32_overflow(self): | ||
| a = testing.shaped_arange((2**32 + 1,), cupy, numpy.float64) | ||
| a = cupy.arange(2**32 + 1, dtype=cupy.float64) |
Collaborator
There was a problem hiding this comment.
Another raw call with float64
ndgrigorian
reviewed
Sep 30, 2026
| # @pytest.mark.thread_unsafe(reason="allocation too large.") | ||
| def test_argmin_int32_overflow(self): | ||
| a = testing.shaped_arange((2**32 + 1,), cupy, numpy.float64) | ||
| a = cupy.arange(2**32 + 1, dtype=cupy.float64) |
This branch has not been deployed
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.
This PR refreshes the vendored third-party test suite under
dpnp/tests/third_party/cupyandcupyx: it syncs the tests with upstream changes, migrates and expands coverage, adapts to newer NumPy/SciPy, and makes the newly added tests portable across dpnp devices (CPU/GPU, with and without native fp64 support). The change is test-only — no library or runtime code is modified.What's included
New and expanded coverage:
cumulative_sum/cumulative_prod(Array API) and batch-scan kernel tests forcumsum/cumprodcupy.linalg.tensordot,shapeargument toreshape, and additional eigenvalue testscupyx.scipy.specialerftests, and additionaltest_search/test_join/test_ndarraycasesFramework and compatibility updates:
test_raw.pyto pytestCorrectness fixes surfaced by the new tests:
rfftn/fftnshape mismatch whensand non-defaultaxesare given togetherDevice portability for the new batch-scan tests:
cumprodtolerance on the float32 path (rtol=1e-5,atol=1e-6); keep the fp64 path strictChecklist