Repository navigation
Conversation
Adds an optional `wgpu::PipelineCache` to `RenderSettings` for use by `Renderer::new_with`. The cache defaults to `None` and is available with the `wgpu` feature. Settings are now Clone rather than Copy because they own the cache handle. The driver's shader compilation dominates the renderer's creation on mobile GPUs: 763 ms on a Mali-G57 MC2 and 1241 ms on an Adreno 610 from a cold start, measured in an Android host with simpleperf (libGLES_mali / libllvm-qgl, not naga). A cache the host persists between runs (`PipelineCache::get_data`) makes that a few milliseconds.
taj-p
force-pushed
the
tajp/pipeline-cache
branch
from
October 5, 2026 19:40
ca904ea to
332a7ba
Compare
LaurenzV
approved these changes
Oct 6, 2026
| /// | ||
| /// Requires [`wgpu::Features::PIPELINE_CACHE`] on the device. | ||
| #[cfg(feature = "wgpu")] | ||
| pub pipeline_cache: Option<wgpu::PipelineCache>, |
Collaborator
There was a problem hiding this comment.
I think having public struct fields gated behind features is usually a bad idea. If someone depends on Vello GPU to only use the WebGL backend, but then some other dependency in the tree activates the wgpu backend as well, their code will fail to compile. I think we should either make the fields in RenderSettings private and turn it into a builder pattern, or instead have new_with take the pipeline cache directly. The second is probably easier for now? Even if it requires updating more call sites.
This branch has not been deployed
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Background
Testing Vello GPU on a Samsung A24 (Mali-G57) and A05s (Adreno 610) sees pipeline compilation take 763ms and 1241ms respectively. This impacts perceived start up performance.
Intent
Use a pipeline cache to push pipeline compilation latency to several milliseconds.
AI Contribution: Codex Astra + Claude Fable