-
Notifications
You must be signed in to change notification settings - Fork 788
[doc][ext] Update HIP/L0 support in p2p doc #15847
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Conversation
Signed-off-by: JackAKirk <[email protected]>
However note that it seems that Level Zero support may require greater clarification/testing: #15841 |
Previously I tested with oneapi standalone dpcppp compiler downloaded from website but now I cloned one of the latest commits from the repository that I know worked in my other system with Nvidia GPUs and tested on that commit in my Intel GPU VM system now it prints out :
On the other hand, the installation test
With Intel OneAPI loaded via
These are devices
I am still experiencing the error message in #15841 and sometimes instead of exception error message it prints out the message "P2P access is not supported by devices, exiting."
|
I think that this probably explains the recent p2p issues: ee97531 I'll try to fix this. |
Related: oneapi-src/unified-runtime#2079 |
This is fixed by oneapi-src/unified-runtime#2246 and #15873 |
These PRs are merged. This means that Level zero support should be fixed and the update to the docs in this PR is correct. Thanks |
The extension document was out of date since it stated that P2P access only worked on CUDA. This updates it to clarify that P2P access is also supported on hip and l0 backends.