Merge master into include-dpctl-tensor#2776
Merged
vlad-perevezentsev merged 19 commits intoinclude-dpctl-tensorfrom Feb 19, 2026
Merged
Merge master into include-dpctl-tensor#2776vlad-perevezentsev merged 19 commits intoinclude-dpctl-tensorfrom
vlad-perevezentsev merged 19 commits intoinclude-dpctl-tensorfrom
Conversation
The PR aligns with recent change in NumPy and adds more clarity on the description of the `file` positional argument on `dpnp.fromfile` documentation.
…#2748) The PR updates GitHub action with testing dpnp conda package.
This PR updates the `.pre-commit-config.yaml` using `pre-commit autoupdate`.
The PR aligns with NumPy 2.4 change and adds support for tuple of integers passed with `axis` argument in `dpnp.trim_zeros` - [x] Have you provided a meaningful PR description? - [x] Have you added a test, reproducer or referred to an issue with a reproducer? - [x] Have you tested your changes locally for CPU and GPU devices? - [x] Have you made sure that new changes do not introduce compiler warnings? - [ ] Have you checked performance impact of proposed changes? - [x] Have you added documentation for your changes, if necessary? - [x] Have you added your changes to the changelog?
Bumps [github/codeql-action](https://github.com/github/codeql-action) from 4.31.11 to 4.32.0.
Bumps [github/codeql-action](https://github.com/github/codeql-action) from 4.32.0 to 4.32.2.
The PR changes implementation of `strides` property in `dpnp.ndarray` to align with NumPy and CuPy and to return bytes displacement in memory (previously and in dpctl it returns elements displacement).
…eginf` (#2754) The PR extends implementation of `dpnp.nan_to_num` function to align with NumPy and CuPy which supports `nan`, `posinf`, and `neginf` keywords as any array through broadcasting. This PR adds handling for a common path where at least one of the keywords has non-scalar value. The path does not assume a dedicated SYCL kernel, instead proposes to rely on implementation through existing python functions. That can be improved in the future if required.
This PR bumps `anaconda-client` version from 1.14.1 to 1.14.0.
Otherwise the upload step in `Conda package` workflow is failing with:
```bash
Traceback (most recent call last):
File "/home/runner/miniconda3/envs/upload/bin/anaconda", line 6, in <module>
from anaconda_cli_base.cli import app
File "/home/runner/miniconda3/envs/upload/lib/python3.13/site-packages/anaconda_cli_base/cli.py", line 229, in <module>
load_registered_subcommands(app)
~~~~~~~~~~~~~~~~~~~~~~~~~~~^^^^^
File "/home/runner/miniconda3/envs/upload/lib/python3.13/site-packages/anaconda_cli_base/plugins.py", line 282, in load_registered_subcommands
subcommand_entry_points = _load_entry_points_for_group(PLUGIN_GROUP_NAME)
File "/home/runner/miniconda3/envs/upload/lib/python3.13/site-packages/anaconda_cli_base/plugins.py", line 57, in _load_entry_points_for_group
module: typer.Typer = entry_point.load()
~~~~~~~~~~~~~~~~^^
File "/home/runner/miniconda3/envs/upload/lib/python3.13/importlib/metadata/__init__.py", line 179, in load
module = import_module(match.group('module'))
File "/home/runner/miniconda3/envs/upload/lib/python3.13/importlib/__init__.py", line 88, in import_module
return _bootstrap._gcd_import(name[level:], package, level)
~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/home/runner/miniconda3/envs/upload/lib/python3.13/site-packages/binstar_client/__init__.py", line 16, in <module>
from pkg_resources import parse_version as pv
ModuleNotFoundError: No module named 'pkg_resources'
```
Due to the latest version `82.0.0` of `setuptools` package is using now
in the env.
This PR introduce the plugin feature to capture and format pytest warnings to process it in internal CI. These changes will not add any overhead to existing pytest scope. This feature can be fully enabled/disabled by env var - `DPNP_INFRA_WARNINGS_ENABLE`
Bumps [github/codeql-action](https://github.com/github/codeql-action) from 4.32.2 to 4.32.3.
…2767) The PR fixes the issue with failing `TestTrimZeros::test_multiple_axes` test when running on a device without fp64 support.
…2765) The PR bumps run dependencies on DPC++ compiler and OneMKL packages in scope of 2026.0 release enabling. Also it includes the workaround for the CMake compiler issue on Windows, which has to be removed once the compiler resolve that.
The PR propose to improve implementation and to use `dpnp.sort` call when - input array has number of dimensions > 1 - input array has previously not supported integer dtype - `axis` keyword is passed (previously not supported) - sequence of `kth` is passed (previously not supported) In case of `ndim > 1` previously the implementation from legacy backend was used, which is significantly slow (see performance comparation below). It used a copy of input data into the shared USM memory and included computations on the host. This PR proposes to reuse `dpnp.sort` for all the above cases. While in case when the legacy implementation is stable and fast (for 1D input array), it will remain, because it relays on `std::nth_element` from OneDPL. The benchmark results were collected on PVC with help of the below code: ```python import dpnp, numpy as np from dpnp.tests.helper import generate_random_numpy_array a = generate_random_numpy_array(10**7, dtype=np.float64, seed_value=117) ia = dpnp.array(a) %timeit x = dpnp.partition(ia, 513); x.sycl_queue.wait() ``` Below tables contains data in case of 1D input array (shape=(10**7,)), where the implementation path was kept the same, plus adding support of missing integer dtypes using fallback on the sort function: | Implementation | int32 | uint32 | int64 | uint64 | float32 | float64 | complex64 | complex128 | |--------|--------|--------|--------|--------|--------|--------|--------|--------| | old (legacy backend) | 7.46 ms | not supported | 9.46 ms | not supported | 7.39 ms | 8.92 ms | 10.9 ms | 21.2 ms | | new (backend + sort) | 7.34 ms | 10.8 ms | 9.48 ms | 12.5 ms | 7.37 ms | 8.89 ms | 11 ms | 21.2 ms | The following code was used for 2D input array with shape=(10**4, 10**4): ```python import dpnp, numpy as np from dpnp.tests.helper import generate_random_numpy_array a = generate_random_numpy_array((10**4, 10**4), dtype=np.float64, seed_value=117) ia = dpnp.array(a) %timeit x = dpnp.partition(ia, 1513); x.sycl_queue.wait() ``` In that case the new implementation is fully based on the sort call: | Implementation | int32 | int64 | float32 | float64 | complex64 | complex128 | |--------|--------|--------|--------|--------|--------|--------| | old (legacy backend) | 6.4 s | 6.89 s | 7.36 s | 7.66 s | 8.61 s | 10 s | | new (sort) | 57.4 ms | 64.7 ms | 62.2 ms | 68 ms | 77 ms | 151 ms |
There is no real dependency on `onemkl-sycl-stats` package. DPNP doesn't consume any function from it. The PR drops the unnecessary run dependency on this oneMKL conda package.
The PR updates CMakeLists.txt to pull pybind11 `3.0.2` up from `3.0.1`.
On Linux, the DPC++ compiler headers are automatically treated as system headers. But it is not the case on Windows. Due to that, there is a ton of deprecation warnings generated inside the compiler headers when building dpnp extensions on Windows. This PR updates CMake files of the extensions to add include of DPC++ compiler headers explicitly and to mark the compiler and dpctl headers as system one to suppress the warning inside them.
vlad-perevezentsev
requested review from
antonwolfy and
ndgrigorian
as code owners
Contributor
|
View rendered docs @ https://intelpython.github.io/dpnp/index.html |
Contributor
|
Array API standard conformance tests for dpnp=0.20.0dev3=py313h509198e_6 ran successfully. |
ndgrigorian
approved these changes
Feb 19, 2026
vlad-perevezentsev
merged commit 464ddc1
into
include-dpctl-tensor
87 of 92 checks passed
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
4 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 synchronizes the
include-dpctl-tensorbranch with the currentmasterbranch