[Impeller] Use Wang's formula for quad/cubic subdivision. #52079
Conversation
This file contains 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
jonahwilliams
commented
Apr 13, 2024
| @@ -78,30 +78,6 @@ TEST(TessellatorTest, TessellatorBuilderReturnsCorrectResultStatus) { | |||
|
|
|||
| ASSERT_EQ(result, Tessellator::Result::kInputError); | |||
| } | |||
|
|
|||
| // More than uint16 points, odd fill mode. | |||
| { | |||
There was a problem hiding this comment.
This is testing a case that no longer matters since we don't use the tessellator. i suspect it broke because we're now generating fewer line segments, since our cubic division isn't as cranky.
As a follow up I plan to delete most of this code.
|
Note: if you zoom in to the remaining overhead, it can be seen that most of it is temporary allocation related:
|
jonahwilliams
changed the title
jonahwilliams
changed the title
|
Golden file changes have been found for this pull request. Click here to view and triage (e.g. because this is an intentional change). If you are still iterating on this change and are not ready to resolve the images on the Flutter Gold dashboard, consider marking this PR as a draft pull request above. You will still be able to view image results on the dashboard, commenting will be silenced, and the check will not try to resolve itself until marked ready for review. |
chinmaygarde
approved these changes
Apr 14, 2024
impeller/geometry/wangs_formula.cc
Outdated
| return std::sqrt(nn.x + nn.y); | ||
| } | ||
|
|
||
| static inline Point Max(Point a, Point b) { |
There was a problem hiding this comment.
Don't we already have Point::Max and Point::GetLength?
There was a problem hiding this comment.
Switched to Point::Max. The length measurement is different however.
|
None of the golden diffs are unexpected either. |
|
Golden file changes are available for triage from new commit, Click here to view. |
jonahwilliams
added
the
autosubmit
|
Part of flutter/flutter#143077 Results: About 5x faster cubics, 2-3x fast quads. |
engine-flutter-autoroll
added a commit
to engine-flutter-autoroll/flutter
that referenced
this pull request
Apr 15, 2024
auto-submit bot
pushed a commit
to flutter/flutter
that referenced
this pull request
Apr 16, 2024
…146790) flutter/engine@07ae93c...503e7e8 2024-04-15 98614782+auto-submit[bot]@users.noreply.github.com Reverts "[Impeller] moved to bgra10_xr (#52019)" (flutter/engine#52140) 2024-04-15 [email protected] Roll Skia from d27e288efd54 to 0b5df0201734 (1 revision) (flutter/engine#52132) 2024-04-15 [email protected] Roll Skia from 0fe107da5a4e to d27e288efd54 (6 revisions) (flutter/engine#52130) 2024-04-15 [email protected] [Impeller] moved to bgra10_xr (flutter/engine#52019) 2024-04-15 [email protected] Roll Skia from 9e20a146c024 to 0fe107da5a4e (33 revisions) (flutter/engine#52129) 2024-04-15 [email protected] [Impeller] Use Wang's formula for quad/cubic subdivision. (flutter/engine#52079) 2024-04-15 [email protected] [Impeller] organize shaders a bit, make filter shaders use same vertex source. (flutter/engine#52113) 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],[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://issues.skia.org/issues/new?component=1389291&template=1850622 Documentation for the AutoRoller is here: https://skia.googlesource.com/buildbot/+doc/main/autoroll/README.md
gilnobrega
pushed a commit
to gilnobrega/flutter
that referenced
this pull request
Apr 22, 2024
…lutter#146790) flutter/engine@07ae93c...503e7e8 2024-04-15 98614782+auto-submit[bot]@users.noreply.github.com Reverts "[Impeller] moved to bgra10_xr (flutter#52019)" (flutter/engine#52140) 2024-04-15 [email protected] Roll Skia from d27e288efd54 to 0b5df0201734 (1 revision) (flutter/engine#52132) 2024-04-15 [email protected] Roll Skia from 0fe107da5a4e to d27e288efd54 (6 revisions) (flutter/engine#52130) 2024-04-15 [email protected] [Impeller] moved to bgra10_xr (flutter/engine#52019) 2024-04-15 [email protected] Roll Skia from 9e20a146c024 to 0fe107da5a4e (33 revisions) (flutter/engine#52129) 2024-04-15 [email protected] [Impeller] Use Wang's formula for quad/cubic subdivision. (flutter/engine#52079) 2024-04-15 [email protected] [Impeller] organize shaders a bit, make filter shaders use same vertex source. (flutter/engine#52113) 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],[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://issues.skia.org/issues/new?component=1389291&template=1850622 Documentation for the AutoRoller is here: https://skia.googlesource.com/buildbot/+doc/main/autoroll/README.md
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
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.
Use a faster method to compute the number of subdivisions of quadradic and cubic curves. Unlike the adaptive subdivison we use this may give us more line segments around straight-ish parts of a curve ... but who cares! the rasterizer can handle it. This is enough to lop off about 2ms from the animated tessellation benchmark (5.5 to 3.5 on a Pixel 7), and there are some additional savings we can book by removing some of the incidental allocations. This part will be done in a follow up (#52078).
Before
After
From looking at the proproptions I think this is in the 3x-5x faster range.
Code originally taken from https://github.com/google/skia/blob/main/src/gpu/tessellate/WangsFormula.h