Skip to content

Conversation

@sommerlukas
Copy link
Contributor

Full cleanup for fusion placeholder commands is handled by specialized scheduler entry points (start_fusion, cleanUPCmdFusion). To avoid interference with that mechanism, fusion commands should not be added to the deferred cleanup list.

@sommerlukas sommerlukas self-assigned this Oct 19, 2023
@sommerlukas sommerlukas requested a review from a team as a code owner October 19, 2023 13:24
@sommerlukas sommerlukas temporarily deployed to WindowsCILock October 19, 2023 13:26 — with GitHub Actions Inactive
@sommerlukas sommerlukas temporarily deployed to WindowsCILock October 19, 2023 13:46 — with GitHub Actions Inactive
@sommerlukas sommerlukas force-pushed the no-deferred-fusion-cleanup branch from 30b9898 to e9a1dab Compare October 19, 2023 15:04
@sommerlukas sommerlukas temporarily deployed to WindowsCILock October 19, 2023 15:07 — with GitHub Actions Inactive
@sommerlukas sommerlukas temporarily deployed to WindowsCILock October 19, 2023 15:41 — with GitHub Actions Inactive
@sommerlukas
Copy link
Contributor Author

The errors on CUDA CI seems unrelated to this PR:

# .---command stderr------------
# | 
# | UR CUDA ERROR:
# | 	Value:           101
# | 	Name:            CUDA_ERROR_INVALID_DEVICE
# | 	Description:     invalid device ordinal
# | 	Function:        operator()
# | 	Source Location: /__w/llvm/llvm/build/_deps/unified-runtime-src/source/adapters/cuda/platform.cpp:75
# | 
# | terminate called after throwing an instance of 'sycl::_V1::runtime_error'
# |   what():  No device of requested type available. -1 (PI_ERROR_DEVICE_NOT_FOUND)
# `-----------------------------
# error: command failed with exit status: -6

@sommerlukas
Copy link
Contributor Author

@sergey-semenov Friendly ping.

Copy link
Contributor

@sergey-semenov sergey-semenov left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Please, update the PR with the latest changes from the sycl branch since the test results are quite stale.

Full cleanup for fusion placeholder commands is handled by specialized
scheduler entry points (`start_fusion`, `cleanUPCmdFusion`). To avoid
interference with that mechanism, fusion commands should not be added to
the deferred cleanup list.

Signed-off-by: Lukas Sommer <[email protected]>
@sommerlukas
Copy link
Contributor Author

Updated the branch as requested by @sergey-semenov.

Test failures are unrelated to this PR:

@intel/llvm-gatekeepers Can someone merge this please?

@steffenlarsen steffenlarsen merged commit 1580c39 into intel:sycl Nov 30, 2023
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.

3 participants