Uh oh!
There was an error while loading. Please reload this page.
Fix to use resident device stream in IPC peer access tests - #1568
Conversation
Auto-sync is disabled for ready for review pull requests in this repository. Workflows must be run manually. Contributors can view more details about this message here. |
rwgk
commented
Feb 3, 2026
/ok to test |
|
| # Test 1: Buffer accessible from resident device (dev1) - should always work | ||
| dev1 = Device(1) | ||
| dev1.set_current() | ||
| dev1_stream = dev1.create_stream() |
There was a problem hiding this comment.
This stream is used for test2, can we move the creation underneath the test 2 comment?
| dev1 = Device(1) | ||
| dev1.set_current() | ||
| dev1_stream = dev1.create_stream() | ||
| PatternGen(dev1, NBYTES).verify_buffer(buffer, seed=False) |
There was a problem hiding this comment.
Just a note that PatternGen uses unified / managed memory underneath it where I suspect this may give us some pain / troubles on Windows?
There was a problem hiding this comment.
Side remark: I'm working on getting rid of that distraction under #1565
In the meantime we discovered that the root cause is much deeper.
I saw @testcupy closed the internal PR. Closing this copy, too.
Removed preview folders for the following PRs: - PR #1568
Description
This resolves failures in QA environments (nvbug 5821337; internal PR 242).