Skip to content

Conversation

steffenlarsen
Copy link
Contributor

This commit adds functions for handling inter-process communication (IPC) of device USM memory allocations. Support for these functions are implemented for both level zero adapters and the CUDA adapter.

This commit adds functions for handling inter-process communication
(IPC) of device USM memory allocations. Support for these functions are
implemented for both level zero adapters and the CUDA adapter.

Signed-off-by: Larsen, Steffen <[email protected]>
@steffenlarsen
Copy link
Contributor Author

Draft SYCL extension building on this: #20018

Signed-off-by: Larsen, Steffen <[email protected]>
Signed-off-by: Larsen, Steffen <[email protected]>
Signed-off-by: Larsen, Steffen <[email protected]>
Signed-off-by: Larsen, Steffen <[email protected]>
@steffenlarsen
Copy link
Contributor Author

After some user feedback, the current API should reflect their needs in relation to #20018.

Friendly ping - @intel/dpcpp-nativecpu-reviewers @intel/llvm-reviewers-cuda @intel/unified-runtime-reviewers-opencl @intel/unified-runtime-reviewers-level-zero @intel/unified-runtime-reviewers

@steffenlarsen
Copy link
Contributor Author

Friendly ping - @intel/dpcpp-nativecpu-reviewers @intel/llvm-reviewers-cuda @intel/unified-runtime-reviewers-opencl @intel/unified-runtime-reviewers-level-zero @intel/unified-runtime-reviewers

@bratpiorka
Copy link
Contributor

CUDA/OpenCL/HIP + UMF usage LGTM

Signed-off-by: Larsen, Steffen <[email protected]>
Signed-off-by: Larsen, Steffen <[email protected]>
@steffenlarsen
Copy link
Contributor Author

CUDA/OpenCL/HIP + UMF usage LGTM

Thank you, @bratpiorka ! I had some feedback on the SYCL extension which changes the approach a little, primarily dropping the UR struct entirely in favor of using the handle data directly. I will make sure it stabilizes and you can have another look at the changes.

@bratpiorka
Copy link
Contributor

Thank you, @bratpiorka ! I had some feedback on the SYCL extension which changes the approach a little, primarily dropping the UR struct entirely in favor of using the handle data directly. I will make sure it stabilizes and you can have another look at the changes.

yes, I see, great - in my opinion, it looks simpler for the user

@steffenlarsen
Copy link
Contributor Author

@intel/unified-runtime-reviewers-level-zero - Friendly ping.

Signed-off-by: Larsen, Steffen <[email protected]>
@steffenlarsen
Copy link
Contributor Author

There are no longer any active reviewers in @intel/dpcpp-nativecpu-reviewers so I will bypass their approval.

@steffenlarsen steffenlarsen merged commit f120751 into intel:sycl Oct 16, 2025
75 of 86 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

4 participants