This repository was archived by the owner on Feb 25, 2025. It is now read-only.
[Impeller] Fix 1d threadgroup dispatch size, Implement drawPoint geometry computation with compute, fallback to specialized CPU. #42060
Conversation
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
added 30 commits
May 7, 2023 12:34
dnfield
reviewed
May 18, 2023
| layout(local_size_x = 16) in; | ||
|
|
||
| layout(std430) readonly buffer PointData { | ||
| vec2 points[]; |
Contributor
There was a problem hiding this comment.
You should include a uint count here to guard against OOB access.
Contributor
There was a problem hiding this comment.
Oh, I see, you put it as a uniform.
Contributor
There was a problem hiding this comment.
(Consider adding a comment that the count for this is in the uniform)
dnfield
reviewed
May 18, 2023
| point_data; | ||
|
|
||
| layout(std430) writeonly buffer GeometryData { | ||
| vec2 geometry[]; |
Contributor
There was a problem hiding this comment.
Probably want a count here too?
dnfield
reviewed
May 18, 2023
Comment on lines
33
to
34
| // TODO(dnfield): https://github.com/flutter/flutter/issues/112683 | ||
| // We should be able to use length here instead of an extra arrgument. |
Contributor
There was a problem hiding this comment.
I kind of doubt this will ever get fixed, we should probably just remove these.
Contributor
Author
There was a problem hiding this comment.
We can actually ifdef/function constant these out based on a runtime flag if the device supports non-uniform threadgroup sizing.
…nto draw_with_compute
dnfield
approved these changes
May 19, 2023
Contributor
Author
|
I found a problem with how dispatch was being done. We were setting the threadGroup size as if it were the threadsize, which resulted in us doing way more work than necessary. I've now fixed this for one dimension, so that the thread group size is correctly computed as the units of work / maxThreadgroupSize. This is still broken for 2D. fyi @dnfield |
jonahwilliams
changed the title
jonahwilliams
commented
May 19, 2023
|
|
||
| // For now, check that the sizes are uniform. | ||
| FML_DCHECK(grid_size == thread_group_size); | ||
| // FML_DCHECK(grid_size == thread_group_size); |
Contributor
Author
There was a problem hiding this comment.
I think we should rethink this API. What would make sense to me is 1. specifying the units of work in a grid size.
Maybe there are cases where we need to control threadgroup size? not sure
dnfield
reviewed
May 19, 2023
| } | ||
|
|
||
| auto size = MTLSizeMake(width, height, 1); | ||
| [encoder dispatchThreadgroups:size threadsPerThreadgroup:size]; |
Contributor
There was a problem hiding this comment.
This is ok but we should really fix this in a separate PR
dnfield
approved these changes
May 19, 2023
jonahwilliams
added
the
autosubmit
engine-flutter-autoroll
added a commit
to engine-flutter-autoroll/flutter
that referenced
this pull request
May 19, 2023
…Point geometry computation with compute, fallback to specialized CPU. (flutter/engine#42060)
auto-submit bot
pushed a commit
to flutter/flutter
that referenced
this pull request
May 19, 2023
…127224) flutter/engine@2b14f8a...3267fa2 2023-05-19 [email protected] [Impeller] Limit blur kernel to 1000x1000 pixels (flutter/engine#42154) 2023-05-19 [email protected] Roll Fuchsia Mac SDK from IGeTvUK2wkajmQQ2k... to pwdDQgM88sqLmZczj... (flutter/engine#42163) 2023-05-19 [email protected] Roll Skia from 0a63dbe8cd15 to 7202b405f061 (1 revision) (flutter/engine#42162) 2023-05-19 [email protected] [Impeller] Fix 1d threadgroup dispatch size, Implement drawPoint geometry computation with compute, fallback to specialized CPU. (flutter/engine#42060) Also rolling transitive DEPS: fuchsia/sdk/core/mac-amd64 from IGeTvUK2wkaj to pwdDQgM88sqL If this roll has caused a breakage, revert this CL and stop the roller using the controls here: https://autoroll.skia.org/r/flutter-engine-flutter-autoroll Please CC [email protected],[email protected] on the revert to ensure that a human is aware of the problem. To file a bug in Flutter: https://github.com/flutter/flutter/issues/new/choose To report a problem with the AutoRoller itself, please file a bug: https://bugs.chromium.org/p/skia/issues/entry?template=Autoroller+Bug Documentation for the AutoRoller is here: https://skia.googlesource.com/buildbot/+doc/main/autoroll/README.md
CaseyHillers
pushed a commit
to CaseyHillers/flutter
that referenced
this pull request
May 24, 2023
…lutter#127224) flutter/engine@2b14f8a...3267fa2 2023-05-19 [email protected] [Impeller] Limit blur kernel to 1000x1000 pixels (flutter/engine#42154) 2023-05-19 [email protected] Roll Fuchsia Mac SDK from IGeTvUK2wkajmQQ2k... to pwdDQgM88sqLmZczj... (flutter/engine#42163) 2023-05-19 [email protected] Roll Skia from 0a63dbe8cd15 to 7202b405f061 (1 revision) (flutter/engine#42162) 2023-05-19 [email protected] [Impeller] Fix 1d threadgroup dispatch size, Implement drawPoint geometry computation with compute, fallback to specialized CPU. (flutter/engine#42060) Also rolling transitive DEPS: fuchsia/sdk/core/mac-amd64 from IGeTvUK2wkaj to pwdDQgM88sqL If this roll has caused a breakage, revert this CL and stop the roller using the controls here: https://autoroll.skia.org/r/flutter-engine-flutter-autoroll Please CC [email protected],[email protected] on the revert to ensure that a human is aware of the problem. To file a bug in Flutter: https://github.com/flutter/flutter/issues/new/choose To report a problem with the AutoRoller itself, please file a bug: https://bugs.chromium.org/p/skia/issues/entry?template=Autoroller+Bug Documentation for the AutoRoller is here: https://skia.googlesource.com/buildbot/+doc/main/autoroll/README.md
Sign up for free
to subscribe to this conversation on GitHub.
Already have an account?
Sign in.
3 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.
Compute shader implementation of #41803
I found this to be generally more straightforward, and we can revisit the vertex shader approach with GLES.
Benchmark is pending here: flutter/flutter#126728
Note: rendering behavior is identical even though this is now a single draw call. This is becuase strokes have overdraw prevention enabled by default, while this new contents does not