-
Notifications
You must be signed in to change notification settings - Fork 1.2k
Move subgroup size handling in line with spec #8609
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
Conversation
5692ec1 to
6060a05
Compare
| /// | ||
| /// See <https://gpuweb.github.io/gpuweb/#gpuadapterinfo> | ||
| /// where you can always use these values on all devices. | ||
| pub const MAXIMUM_SUBGROUP_MAX_SIZE: u32 = 128; |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Q: Is it possible to use more than this on some machines? If so, do we have any follow-up work for permitting larger groups?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
No - the max observed subgroup size in the wild is 128 on qualcomm chips (and WARP). Similarly WARP supports the same range (4 to 128) so you can test with the full range of subgroups in the wild.
ErichDonGubler
left a comment
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
LGTM for merging. I have an open question, but I'm trusting this will get resolved either at or after merge.
Co-authored-by: JMS55 <[email protected]>
1bfdfef to
ac504eb
Compare
Connections
Closes #8346
Description
Moved things over, as well as fix a few minor things in wgpu-info.
I have updated the defaults to the values specified by the WebGPU spec.
Testing
CI
Squash or Rebase?
Squash