provide optimized APIs accepting GPU device handles or ordinals #66

Open

Labels

opened

on May 22, 2018

The current APIs have an extra overhead due to retrieving the GPU ordinal.
That could be optimized away if the users directly passed the GPU handle/ordinal corresponding to the stream.

to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Assignees

No one assigned

Labels

Type

No type

Projects

No projects

Milestone

No milestone

Relationships

None yet

Development

No branches or pull requests