Skip to content
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

Revert "Update Kokkos to 4.2 in the SYCL build" #993

Merged
merged 1 commit into from
Dec 29, 2023

Conversation

aprokop
Copy link
Contributor

@aprokop aprokop commented Dec 23, 2023

This reverts commit c47f38f.

SYCL has been broken since #973. This PR is to try to make the CI pass again. We either downgrade, or figure out what's happening.

@aprokop
Copy link
Contributor Author

aprokop commented Dec 23, 2023

So, SYCL with Kokkos 4.1 passes. With 4.2 does not:

PI CUDA ERROR:
	Value:           700
	Name:            CUDA_ERROR_ILLEGAL_ADDRESS
	Description:     an illegal memory access was encountered
	Function:        cuda_piextUSMEnqueueMemcpy
	Source Location: /root/intel-llvm-mirror/sycl/plugins/cuda/pi_cuda.cpp:5023


PI CUDA ERROR:
	Value:           700
	Name:            CUDA_ERROR_ILLEGAL_ADDRESS
	Description:     an illegal memory access was encountered
	Function:        operator()
	Source Location: /root/intel-llvm-mirror/sycl/plugins/cuda/pi_cuda.cpp:2553

terminate called after throwing an instance of 'std::runtime_error'
  what():  There was a synchronous SYCL error:
Native API failed. Native API returns: -999 (Unknown PI error) -999 (Unknown PI error)
unknown location(0): fatal error: in "half_traversal<Kokkos__Device<Kokkos__Experimental__SYCL_ Kokkos__Experimental__SYCLDeviceUSMSpace>>": signal: SIGABRT (application abort requested)
/var/jenkins/workspace/ArborX_PR-969/test/tstDetailsHalfTraversal.cpp(60): last checkpoint: "half_traversal" test entry

Should try 4.2 with newer Intel Compiler version, like 2024.0.2.

@aprokop aprokop added the build Build and installation label Dec 23, 2023
@masterleinad
Copy link
Collaborator

I'll have a look next week, we might need to adjust the maximum workgroup sizes for reductions with Kokkos+SYCL+Cuda.

@masterleinad
Copy link
Collaborator

Oh, actually we should just enforce running on an A100 instead of Volta70.

@masterleinad masterleinad mentioned this pull request Dec 23, 2023
@aprokop aprokop merged commit ce20b96 into arborx:master Dec 29, 2023
1 check passed
@aprokop aprokop deleted the fix_sycl_build branch December 29, 2023 20:22
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
build Build and installation
Projects
None yet
Development

Successfully merging this pull request may close these issues.

2 participants