Skip to content

Conversation

@LaurenzV
Copy link
Contributor

I also measured this using the Blend2D benchmark suite and it doesn't result in in any noticeable slowdown.

@LaurenzV
Copy link
Contributor Author

LaurenzV commented Sep 22, 2025

I should note it's a bit unfortunate the size of a command increases, but I think if we really want to optimize for that there are other avanues we could explore first (for example the paint field is currently very big since it currently stores the f32 version and u8 version of a color inline. I could also put it behind a feature flag somehow s that at least vello_hybrid is not affected.

@LaurenzV LaurenzV force-pushed the laurenz/inline_mask branch from 9cec693 to eea9eb4 Compare November 15, 2025 13:22
@LaurenzV LaurenzV requested a review from taj-p November 15, 2025 13:38
@LaurenzV
Copy link
Contributor Author

I've rebased this onto main now. I also reran the Blend2D benchmark harness, and there does seem to be a slight slowdown in some cases, but nothing worrying (around 1-2%, in extreme cases 3%). It is a bit unfortunate that every command stores an Option now, so maybe there is something better we can do here, but it's probably also not a huge deal either?

@taj-p
Copy link
Contributor

taj-p commented Nov 21, 2025

Sorry for the delay. I'll review this over the next couple days

@LaurenzV
Copy link
Contributor Author

All good, it can wait until after your vacation! Enjoy it!

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

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

3 participants