Feature/elementwise functions out keyword

[vtavana]

Adding out keyword for elementwise functions

• 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?
• If this PR is a work in progress, are you opening the PR as a draft?

[github-actions]

View rendered docs @ https://intelpython.github.io/dpctl/pulls/1209/index.html

[github-actions]

Array API standard conformance tests for dpctl=0.14.3dev1=py310h76be34b_98 ran successfully.
Passed: 220
Failed: 780
Skipped: 116

[github-actions]

Array API standard conformance tests for dpctl=0.14.3dev1=py310h76be34b_108 ran successfully.
Passed: 226
Failed: 774
Skipped: 116

[oleksandr-pavlyk]

In [1]: import dpctl.tensor as dpt, numpy as np

In [2]: m = dpt.ones((10, 10), dtype="f4")

In [3]: v = dpt.ones(10, dtype="i4")

In [4]: r = dpt.empty_like(m)

In [5]: dpt.add(m, v, out=r)
---------------------------------------------------------------------------
TypeError                                 Traceback (most recent call last)
Input In [5], in <cell line: 1>()
----> 1 dpt.add(m, v, out=r)

File ~/repos/dpctl/dpctl/tensor/_elementwise_common.py:406, in BinaryElementwiseFunc.__call__(self, o1, o2, out, order)
    401     raise TypeError(
    402         f"output array must be of usm_ndarray type, got {type(out)}"
    403     )
    405 if out.shape != o1_shape or out.shape != o2_shape:
--> 406     raise TypeError(
    407         "The shape of input and output arrays are inconsistent."
    408         f"Expected output shape is {o1_shape}, got {out.shape}"
    409     )
    411 if ti._array_overlap(o1, out) or ti._array_overlap(o2, out):
    412     raise TypeError("Input and output arrays have memory overlap")

TypeError: The shape of input and output arrays are inconsistent.Expected output shape is (10, 10), got (10, 10)

[github-actions]

Array API standard conformance tests for dpctl=0.14.3dev1=py310h76be34b_108 ran successfully.
Passed: 227
Failed: 773
Skipped: 116

[oleksandr-pavlyk]

Changes in this PR lead to a drop in coverage (84.036% in the target branch vs. 83.897% in this branch). Since your changes only modify _elementwise_common.py, please use pytest-cov to identify gaps and add required tests.

[vtavana]

Changes in this PR lead to a drop in coverage (84.036% in the target branch vs. 83.897% in this branch). Since your changes only modify _elementwise_common.py, please use pytest-cov to identify gaps and add required tests.

I added new tests and now the coverage for _elementwise_common.py is 91% compared to 89%. All the lines that are added in this PR are tested now.

after adding new tests
dpctl/tensor/_elementwise_common.py 300 26 176 15 91% 83, 151, 161, 185-186, 229, 239, 243, 254, 259, 277, 315, 318, 341, 361, 372-373, 390, 475-477, 487, 517-519, 529

before adding new tests:
dpctl/tensor/_elementwise_common.py 300 30 176 20 89% 83, 95, 119->124, 151, 161, 185-186, 229, 239, 243, 254, 259, 277, 315, 318, 341, 361, 372-373, 390, 459, 475-477, 487, 496, 517-519, 529, 538

[github-actions]

Array API standard conformance tests for dpctl=0.14.3dev1=py310h76be34b_110 ran successfully.
Passed: 227
Failed: 773
Skipped: 116

[oleksandr-pavlyk]

Tests fail on my WSL where GPU does not support fp64.

Specifically test_add_dtype_matrix where NumPy returns "double" and test_add_dtype_error which attempts to use "f8" regardless target device capabilities.

---

UPD: I pushed fixes for this and added tests to increase coverage.

[github-actions]

Array API standard conformance tests for dpctl=0.14.3dev1=py310h76be34b_111 ran successfully.
Passed: 228
Failed: 772
Skipped: 116

[github-actions]

Array API standard conformance tests for dpctl=0.14.3dev1=py310h76be34b_112 ran successfully.
Passed: 227
Failed: 773
Skipped: 116

[oleksandr-pavlyk]

I think out should not be allowed to be a positional argument in array API. Could you please change it to a keyword argument in the test suite?

[vtavana]

sure!

[github-actions]

Array API standard conformance tests for dpctl=0.14.3dev1=py310h76be34b_113 ran successfully.
Passed: 228
Failed: 772
Skipped: 116

[oleksandr-pavlyk]

Thank you @vtavana ! Please merge when ready

[github-actions]

Deleted rendered PR docs from intelpython.github.com/dpctl, latest should be updated shortly. 🤞