Fix for crash reported in gh-1654

oleksandr-pavlyk commented 1 month ago

Closes gh-1654

The reason behind the crash was out of bound access to shared local memory accessor.

This also fixes a crash in dpt.sort on CUDA device for sorting of 256 elements of floating point numbers.

[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?
[x] 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[bot] commented 1 month ago

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

github-actions[bot] commented 1 month ago

Array API standard conformance tests for dpctl=0.17.0dev0=py310h15de555_336 ran successfully. Passed: 888 Failed: 17 Skipped: 91

coveralls commented 1 month ago

coverage: 87.948%. remained the same when pulling 58dfb5c6fcbcc75b4b311e2294b1f891bc56d397 on fix-for-crash-gh-1654 into 8b42313b5d496fc1cf2fd603139aead8b1924643 on master.

oleksandr-pavlyk commented 1 month ago

The example that was crashing on CUDA device:

import dpctl.tensor as dpt
x = dpt.arange(256, dtype="i4", device="cuda:gpu")
dpt.sort(x, descending=True) # crash

IntelPython / dpctl

Fix for crash reported in gh-1654 #1676