Rietveld Code Review Tool
Help | Bug tracker | Discussion group | Source code | Sign in
(1677)

Issue 6160046: Added cache to gpu AA clipping to accelerate (Closed)

Can't Edit
Can't Publish+Mail
Start Review
Created:
12 years, 4 months ago by robertphillips
Modified:
12 years, 4 months ago
Reviewers:
reed1, TomH
CC:
skia-review_googlegroups.com
Base URL:
http://skia.googlecode.com/svn/trunk/
Visibility:
Public.

Description

Unsurprisingly, the stencil clip path relies heavily on the stencil buffer "caching" the last clip. This patch adds the same capability to the AA clip path so the clip mask isn't constantly being re-rendered. A future question is: would having more than 1 cached mask ever be worthwhile? I suspect not.

Patch Set 1 #

Unified diffs Side-by-side diffs Delta from patch set Stats (+69 lines, -4 lines) Patch
M src/gpu/GrClipMaskManager.h View 5 chunks +57 lines, -2 lines 0 comments Download
M src/gpu/GrClipMaskManager.cpp View 4 chunks +12 lines, -2 lines 0 comments Download

Messages

Total messages: 4
robertphillips
12 years, 4 months ago (2012-05-02 19:13:13 UTC) #1
TomH
LGTM. For "relies heavily" I assume you mean "is absurdly slow without"?
12 years, 4 months ago (2012-05-02 19:23:11 UTC) #2
robertphillips
Committed as r3828 and yes, rendering w/o at least 1 level of cache is unbelievably ...
12 years, 4 months ago (2012-05-02 19:34:20 UTC) #3
reed1
12 years, 4 months ago (2012-05-02 19:41:06 UTC) #4
FYI --

The raster side keeps a stack-cache of its rendered clips (hard and soft), as
the API is already in stack order...

save
    clip...
    ...
    save
        clip...
        ...
    restore
    ...
restore
Sign in to reply to this message.

Powered by Google App Engine
RSS Feeds Recent Issues | This issue
This is Rietveld f62528b