xla
Make from_dlpack handle cuda synchronization implicitly for input tensors that have dlpack and __dlpack_device__ attributes.
#7125

Merged

Make from_dlpack handle cuda synchronization implicitly for input tensors that have dlpack and __dlpack_device__ attributes. #7125

vanbasten23 merged 6 commits into master from xiowei/use_producer_dlpack_device

vanbasten23 changed the title ~~Handle cuda synchronize implicitly~~ Make from_dlpack handle cuda synchronize implicitly for input tensors that have __dlpack__ and __dlpack_device__ attributes. 2 years ago

vanbasten23 changed the title ~~Make from_dlpack handle cuda synchronize implicitly for input tensors that have __dlpack__ and __dlpack_device__ attributes.~~ Make from_dlpack handle cuda synchronization implicitly for input tensors that have __dlpack__ and __dlpack_device__ attributes. 2 years ago

vanbasten23 requested a review from

ysiraichi 2 years ago

vanbasten23 requested a review from

JackCaoG 2 years ago

vanbasten23 requested a review from

will-cromar 2 years ago

vanbasten23 marked this pull request as ready for review 2 years ago

JackCaoG commented on 2024-05-28

leverage __dlpack__ from the producer

63ffffe6

updated the test

bd7d2655

Add a fialing test test_dlpack_xla_to_pytorch_cuda_protocol_conversion

a5e43534

Revert "Add a fialing test test_dlpack_xla_to_pytorch_cuda_protocol_c…

c51be878

remove unwanted

ec2d3963

rename test and run linter

c4898511

vanbasten23 force pushed from d42d4536 to c4898511 2 years ago

vanbasten23 requested a review from

JackCaoG 2 years ago

JackCaoG commented on 2024-05-30

JackCaoG approved these changes on 2024-05-30

vanbasten23 merged daada224 into master 2 years ago

Reviewers

JackCaoG

ysiraichi

will-cromar

Assignees

No one assigned

Labels

None yet

Milestone

No milestone

xla Make from_dlpack handle cuda synchronization implicitly for input tensors that have __dlpack__ and __dlpack_device__ attributes. #7125 Merged

Make from_dlpack handle cuda synchronization implicitly for input tensors that have __dlpack__ and __dlpack_device__ attributes. #7125

xla
Make from_dlpack handle cuda synchronization implicitly for input tensors that have dlpack and __dlpack_device__ attributes.
#7125

Merged

Make from_dlpack handle cuda synchronization implicitly for input tensors that have dlpack and __dlpack_device__ attributes. #7125