Supporting order arg to asnumpy#2980
Open
abagusetty wants to merge 7 commits into
Open
Conversation
dpnp.asnumpy and dpnp.ndarray.asnumpy accepted an order argument but dropped it for dpnp_array and usm_ndarray inputs. The underlying _copy_to_numpy rebuilt the NumPy array from the source strides, so a non-contiguous source was returned with a non-contiguous layout even when order='C' was requested, diverging from NumPy/CuPy. Thread order through the full conversion chain: - _copy_to_numpy(ary, order='K'): apply layout via np.asarray - dpt.asnumpy(usm_ary, order='K'): forward order - dpnp_array.asnumpy(order='C'): forward order - dpnp.asnumpy(a, order='C'): pass order to both array branches Public APIs default to 'C' to match NumPy/CuPy; internal helpers default to 'K' to preserve the existing stride-keeping behavior of their direct callers (to_numpy, _ctors, _print). Add TestAsNumpy covering iface/method/usm_ndarray paths and default semantics.
abagusetty
requested review from
antonwolfy,
ndgrigorian and
vlad-perevezentsev
as code owners
|
Can one of the admins verify this patch? |
Collaborator
|
coverage: 78.071% (-0.002%) from 78.073% — abagusetty:fix-asnumpy-order-2884 into IntelPython:master |
antonwolfy
reviewed
Jun 30, 2026
| Args: | ||
| usm_ary (usm_ndarray): | ||
| Input array | ||
| order (``"C"``, ``"F"``, ``"A"``, ``"K"``): |
Contributor
There was a problem hiding this comment.
Suggested change
| order (``"C"``, ``"F"``, ``"A"``, ``"K"``): | |
| order (``"C"``, ``"F"``, ``"A"``, ``"K"``, optional): |
Contributor
There was a problem hiding this comment.
I believe None value is also support for tensor.asnumpy
| if ary.size == 0: | ||
| # no data needs to be copied for zero sized array | ||
| return np.ndarray(ary.shape, dtype=ary.dtype) | ||
| return np.ndarray(ary.shape, dtype=ary.dtype, order=order) |
Contributor
There was a problem hiding this comment.
I wonder if it is allowed to pass K and A values here, because based on the documentation it's limited to C and F only:
{‘C’, ‘F’}, optional
Contributor
There was a problem hiding this comment.
Any size == 0 array reports C_CONTIGUOUS = True and F_CONTIGUOUS = True regardless of order.
It'd make sense to leave as original:
Suggested change
| return np.ndarray(ary.shape, dtype=ary.dtype, order=order) | |
| return np.ndarray(ary.shape, dtype=ary.dtype) |
but place a comment
| def test_method_order_k_keeps_strides(self): | ||
| # explicit "K" keeps the strides of the source as closely as possible | ||
| a = self._non_c_contiguous_array() | ||
| result = a.asnumpy(order="K") |
Contributor
There was a problem hiding this comment.
missing test for order="A" and order=None
| a = self._non_c_contiguous_array() | ||
| result = a.asnumpy(order="K") | ||
| assert not result.flags["C_CONTIGUOUS"] | ||
|
|
Contributor
There was a problem hiding this comment.
Messing test with zero-sized ndarray
Co-authored-by: Anton <100830759+antonwolfy@users.noreply.github.com>
Contributor
Author
|
@antonwolfy Thanks for taking time. Addressed the above suggestions |
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.
Fixed
dpnp.asnumpyanddpnp.ndarray.asnumpyignoring theorderargFixes: #2568