Skip to content

Move tensor tests from dpctl - #2817

Merged
vlad-perevezentsev merged 363 commits into
include-dpctl-tensorfrom
move_tensor_tests
Apr 10, 2026
Merged

Move tensor tests from dpctl #2817
vlad-perevezentsev merged 363 commits into
include-dpctl-tensorfrom
move_tensor_tests

Conversation

@vlad-perevezentsev

@vlad-perevezentsevvlad-perevezentsev commented Mar 12, 2026•
edited
Loading

Copy link
Copy Markdown
Contributor

This PR moves all tensor-related tests to dpnp/tests/tensor as part of the ongoing migration of tensor functionality from dpctl to dpnp.tensor

Key changes:

  • Relocated 89 tensor tests (elementwise functions, usm_ndarray, and tensor utilities)
  • Updated imports to use dpnp.tensor
  • Included tests in packaging configuration
  • Integrated tensor tests into CI
  • Fixed several issues discovered during migration (dtype expectations, boolean reductions, etc.)
  • Fixed a circular import in _usmarray.py
  • Added SKIP_TENSOR_TESTS env variable to manage the launch of the test scope

In a follow-up PR:

  • Conditional logic will be added to run dpctl_ext/tests only when changes affect the tensor code.
  • Array API tests for tensor will be introduced and executed as a separate CI job.
  • Have you provided a meaningful PR description?
  • Have you added a test, reproducer or referred to an issue with a reproducer?
  • Have you tested your changes locally for CPU and GPU devices?
  • Have you made sure that new changes do not introduce compiler warnings?
  • Have you checked performance impact of proposed changes?
  • Have you added documentation for your changes, if necessary?
  • Have you added your changes to the changelog?

@vlad-perevezentsev
Move ti.cumulative_logsumexp() and reuse it in dpnp
d8c3680
@vlad-perevezentsev
Extend ignore-words-list codespell
a1d9b0a
@vlad-perevezentsev
Move utils rich_comparisons.hpp file
8a87fea
@vlad-perevezentsev
Initialize _tensor_sorting_impl extension and move _radix_sort and py…
43a7a04
…_sort
@vlad-perevezentsev
Move _sort_ascending/descending to dpctl_ext.tensor._tensor_sorting_impl
1f018ec
@vlad-perevezentsev
Move ti.sort() and reuse it in dpnp
b62d836
@vlad-perevezentsev
Move ti.unique_counts() and ti.unique_values() to dpctl_ext.tensor
82d202c
@vlad-perevezentsev
Move _radix_argsort_ascending/descending to _tensor_sorting_impl
893cdc3
@vlad-perevezentsev
Move _argsort_ascending/descending to _tensor_sorting_impl
40c2b84
@vlad-perevezentsev
Move ti.argsort() to dpctl_ext.tensor and reuse it in dpnp
6912311
@vlad-perevezentsev
Move _searchsorted_left/right to _tensor_sorting_impl
88a23a2
@vlad-perevezentsev
Fix bug: wrong shape pointer for positions array
148b6e5
@vlad-perevezentsev
Move ti.searchsorted() and reuse it in dpnp
8b010b0
@vlad-perevezentsev
Move dpt.unique_all() and dpt.unique_inverse() and reuse it in dpnp
9766d34
@vlad-perevezentsev
Move _isin to _tensor_sorting_impl
b8ad5ec
@vlad-perevezentsev
Fix bug: wrong shape pointer for dst array
ce570a5
@vlad-perevezentsev
Move dpt.isin() and reuse it in dpnp
a7c6440
@vlad-perevezentsev
Move _topk to _tensor_sorting_impl
62d19f1
@vlad-perevezentsev
Move dpt.top_k()
56d397d
@vlad-perevezentsev
Initialize _tensor_reductions_impl extension and move _all
ad814fb
@vlad-perevezentsev
Add TODO with incorrect logic in reductions_over_axis.hpp
c4f2496
@vlad-perevezentsev
Move ti.all() to dpctl_ext.tensor and reuse it in dpnp
bd3add0
@vlad-perevezentsev
Move _any to _tensor_reductions_impl
b1953df
@vlad-perevezentsev
Move ti.any()/ti.diff() and reuse them in dpnp
32802d6
@vlad-perevezentsev
Move _min_over_axis/_max_over_axis to _tensor_reductions_impl
1b4e498
@vlad-perevezentsev
Move ti.min()/max() and reuse it in dpnp
aa313ff
@vlad-perevezentsev
Move _argmax/argmin_over_axis to _tensor_reductions_impl
c6d600a
@vlad-perevezentsev
Move ti.argmax()/argmin() to dpctl_ext.tensor and reuse them in dpnp
8ae933d
@vlad-perevezentsev
Move _prod/sum_over_axis to _tensor_reductions_impl
2ec3cc7
@vlad-perevezentsev
Move ti.sum()/prod() to dpctl_ext.tensor and reuse them in dpnp
6b27b1b
antonwolfy
antonwolfy reviewed Apr 8, 2026
View reviewed changes
Comment threaddpnp/tensor/libtensor/include/kernels/elementwise_functions/round.hpp
antonwolfy
antonwolfy reviewed Apr 8, 2026
View reviewed changes
Comment thread.github/workflows/conda-package.yml Outdated
Base automatically changed from move_dpctl_ext_to_dpnp_tensor to include-dpctl-tensorApril 8, 2026 14:57
antonwolfy
antonwolfy previously approved these changes Apr 8, 2026
View reviewed changes

@antonwolfyantonwolfy left a comment

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

In general, LGTM

antonwolfy
antonwolfy reviewed Apr 10, 2026
View reviewed changes
Comment threaddpnp/tests/test_ndarray.py Outdated
antonwolfy
antonwolfy approved these changes Apr 10, 2026
View reviewed changes

@antonwolfyantonwolfy left a comment

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Thank you @vlad-perevezentsev, no more comments from me

@vlad-perevezentsev
vlad-perevezentsev merged commit e1aad07 into include-dpctl-tensorApr 10, 2026
20 of 36 checks passed
@vlad-perevezentsev
vlad-perevezentsev deleted the move_tensor_tests branch April 10, 2026 11:42
@vlad-perevezentsevvlad-perevezentsev mentioned this pull request Apr 14, 2026
Merged
7 tasks
vlad-perevezentsev added a commit that referenced this pull request Apr 15, 2026
@vlad-perevezentsev
Add device-aware output dtype fordpt.round()with boolean input (#…
549e940
…2851)
This PR proposes device-aware output dtype resolution for
`dpnp.tensor.round()` with `boolean` input to handle devices that do not
support `float16`
Boolean support for round() was originally added in #2817
[6f5a792](6f5a792)
to match NumPy behavior where numpy.round(bool) returns float16 rather
than an integral type like int8.
However on devices without fp16 support, returning float16 is not
viable.
The bool type mapping was removed from the round kernel and an
acceptance
function `_acceptance_fn_round` was added to ensure the fallback in
`_find_buf_dtype`
prefers floating-point output over integral types for boolean input
Result :
fp16 devices: round(bool) -> float16 non-fp16 devices: round(bool) -> float32
Sign up for freeto join this conversation on GitHub. Already have an account? Sign in to comment

Reviewers

@ndgrigorianndgrigorianndgrigorian left review comments
@antonwolfyantonwolfyantonwolfy approved these changes

Assignees

@vlad-perevezentsevvlad-perevezentsev

Labels

None yet

Projects

None yet

Milestone

No milestone

Development

Successfully merging this pull request may close these issues.

3 participants

@vlad-perevezentsev@ndgrigorian@antonwolfy