Bitwise elementwise functions

[oleksandr-pavlyk]

Implementation of dpt.bitwise_invert, dpt.bitwise_and, dpt.bitwise_or, dpt.bitwise_xor, dpt.bitwise_left_shift, and dpt.bitwise_right_shift per array API.

• 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/1307/index.html

[coveralls]

coverage: 84.814% (+0.02%) from 84.795% when pulling 9a99003 on bitwise-elementwise-functions into fedc71b on master.

[github-actions]

Array API standard conformance tests for dpctl= ran successfully.
Passed: 584
Failed: 416
Skipped: 119

[github-actions]

Array API standard conformance tests for dpctl=0.14.6dev0=py310h7bf5fec_54 ran successfully.
Passed: 584
Failed: 416
Skipped: 119

[github-actions]

Array API standard conformance tests for dpctl=0.14.6dev0=py310h7bf5fec_70 ran successfully.
Passed: 611
Failed: 389
Skipped: 119

[ndgrigorian]

@oleksandr-pavlyk
Array API tests have caught some undesired behavior:

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

In [2]: x = dpt.asarray([1, 1])

In [3]: y = dpt.asarray(64)

In [4]: dpt.bitwise_right_shift(x, y)
Out[4]: usm_ndarray([1, 1])

In [5]: x_np, y_np = dpt.asnumpy(x), dpt.asnumpy(y)

In [6]: np.right_shift(x_np, y_np)
Out[6]: array([0, 0])

[ndgrigorian]

@oleksandr-pavlyk Array API tests have caught some undesired behavior:

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

In [2]: x = dpt.asarray([1, 1])

In [3]: y = dpt.asarray(64)

In [4]: dpt.bitwise_right_shift(x, y)
Out[4]: usm_ndarray([1, 1])

In [5]: x_np, y_np = dpt.asnumpy(x), dpt.asnumpy(y)

In [6]: np.right_shift(x_np, y_np)
Out[6]: array([0, 0])

And oddly enough:

In [8]: dpt.bitwise_right_shift(dpt.asarray([1, 1]), dpt.asarray(62))
Out[8]: usm_ndarray([0, 0])

In [9]: dpt.bitwise_right_shift(dpt.asarray([1, 1]), dpt.asarray(65))
Out[9]: usm_ndarray([0, 0])

which is the expected result.

[oleksandr-pavlyk]

@ndgrigorian C/C++ spec states that bitshift behavior is undefined if the second argument is negative or not less than total type bitwidth. Additional branching was added ensure compliance with spec.

[github-actions]

Array API standard conformance tests for dpctl=0.14.6dev0=py310h7bf5fec_77 ran successfully.
Passed: 612
Failed: 388
Skipped: 119

[ndgrigorian]

Test is fixed so LGTM!

[github-actions]

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

[github-actions]

Array API standard conformance tests for dpctl=0.14.6dev0=py310h7bf5fec_77 ran successfully.
Passed: 612
Failed: 388
Skipped: 119