Skip to content

Conversation

@mdboom
Copy link
Contributor

@mdboom mdboom commented Sep 23, 2025

This is applying the same trick in #894 to the cybind-generated bindings.

Essentially, by inlining the boolean check in every call, we can avoid the much more expensive call through a C function pointer on every single API call.

@copy-pr-bot
Copy link
Contributor

copy-pr-bot bot commented Sep 23, 2025

Auto-sync is disabled for ready for review pull requests in this repository. Workflows must be run manually.

Contributors can view more details about this message here.

@mdboom
Copy link
Contributor Author

mdboom commented Sep 23, 2025

/ok to test

@github-actions

This comment has been minimized.

@mdboom mdboom enabled auto-merge (squash) September 23, 2025 17:15
@mdboom mdboom merged commit d2978aa into NVIDIA:main Sep 23, 2025
70 checks passed
@github-actions
Copy link

Doc Preview CI
Preview removed because the pull request was closed or merged.

Comment on lines 98 to 99
if __py_nvjitlink_init:
return 0
Copy link
Member

Choose a reason for hiding this comment

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

btw internally we dropped these two lines

@leofang
Copy link
Member

leofang commented Oct 9, 2025

For posterity, this PR was backported as part of #1044.

@leofang leofang added enhancement Any code-related improvements cuda.bindings Everything related to the cuda.bindings module labels Oct 9, 2025
@leofang leofang added this to the cuda-python 13.0.2 & 12.9.3 milestone Oct 9, 2025
@mdboom mdboom deleted the fast-cybind-init branch December 9, 2025 16:13
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

cuda.bindings Everything related to the cuda.bindings module enhancement Any code-related improvements

Projects

None yet

Development

Successfully merging this pull request may close these issues.

3 participants