You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Copy file name to clipboardExpand all lines: CHANGELOG.md
+6Lines changed: 6 additions & 0 deletions
Display the source diff
Display the rich diff
Original file line number
Diff line number
Diff line change
@@ -6,6 +6,7 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
6
6
7
7
## [0.20.0] - MM/DD/2026
8
8
9
+
This release introduces a major architectural change: the Array API-compliant tensor implementation has been migrated from `dpctl.tensor` into `dpnp.tensor`, simplifying maintenance, reducing cross-project dependencies, and allows the tensor implementation to evolve within `dpnp`.
9
10
This release changes the license from `BSD-2-Clause` to `BSD-3-Clause`.
10
11
This release achieves `dpnp` compatibility with Python 3.14 and enables distributing `dpnp` packages with the latest Python version.
11
12
Also, that release drops support for Python 3.9, making Python 3.10 the minimum required version.
@@ -28,6 +29,7 @@ Also, that release drops support for Python 3.9, making Python 3.10 the minimum
28
29
* Added implementation of `dpnp.isin` function [#2595](https://github.com/IntelPython/dpnp/pull/2595)
29
30
* Added implementation of `dpnp.scipy.linalg.lu` (SciPy-compatible) [#2787](https://github.com/IntelPython/dpnp/pull/2787)
30
31
* Added support for ndarray subclassing via `dpnp.ndarray.view` method with `type` parameter [#2815](https://github.com/IntelPython/dpnp/issues/2815)
32
+
* Migrated tensor implementation from `dpctl.tensor` into `dpnp.tensor`, making `dpnp` the primary owner of the Array API-compliant tensor layer [#2856](https://github.com/IntelPython/dpnp/pull/2856)
31
33
32
34
### Changed
33
35
@@ -57,6 +59,8 @@ Also, that release drops support for Python 3.9, making Python 3.10 the minimum
57
59
* Updated QR tests to avoid element-wise comparisons for `raw` and `r` modes [#2785](https://github.com/IntelPython/dpnp/pull/2785)
58
60
* Moved all SYCL kernel functors from `backend/extensions/` to a unified `backend/kernels/` directory hierarchy [#2816](https://github.com/IntelPython/dpnp/pull/2816)
* Disabled `dpnp.tensor` tests by default in `conda build --test` to prevent OOM failures during package testing. Set `SKIP_TENSOR_TESTS=0` to re-enable them on systems with enough memory [#2860](https://github.com/IntelPython/dpnp/pull/2860)
@@ -84,6 +88,8 @@ Also, that release drops support for Python 3.9, making Python 3.10 the minimum
84
88
* Resolved an issue with strides calculation in `dpnp.diagonal` to return correct values for empty diagonals [#2814](https://github.com/IntelPython/dpnp/pull/2814)
85
89
* Fixed test tolerance issues for float16 intermediate precision that became visible when testing against conda-forge's NumPy [#2828](https://github.com/IntelPython/dpnp/pull/2828)
86
90
* Ensured device aware dtype handling in `dpnp.identity` and `dpnp.gradient`[#2835](https://github.com/IntelPython/dpnp/pull/2835)
91
+
* Fixed `dpnp.tensor.round` to use device-aware output dtype for boolean input [#2851](https://github.com/IntelPython/dpnp/pull/2851)
92
+
* Resolved a deadlock in `dpnp.linalg.qr` by releasing the GIL before OneMKL `orgqr` call to prevent host tasks contention [#2850](https://github.com/IntelPython/dpnp/pull/2850)
0 commit comments