Rietveld Code Review Tool
Help | Bug tracker | Discussion group | Source code | Sign in
(195)

Issue 7485043: Refactor PictureRecord optimization & added saveLayer/save/clipR/DBM*/restore/restore optimization (Closed)

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

Description

This CL moves the second optimization over from the filter tool into PictureRecord. This new optimization looks for the pattern: SAVE_LAYER SAVE CLIP_RECT DRAW_BITMAP* RESTORE RESTORE where the saveLayer's paint can be folded into the drawBimap*'s. Additionally some minor refactoring of the optimizations has been performed including: splitting off a merge_savelayer_paint_into_drawbitmp subroutine adding a convert_command_to_noop method adding an array of optimization function pointers adding a generic matching method ('match') Performance numbers are in-line with what was seen in https://codereview.appspot.com/7294049/

Patch Set 1 #

Patch Set 2 : Fixed comments #

Patch Set 3 : Hid new optimization behind SK_IGNORE_PICTURE_RECORD_SAVE_LAYER_OPT flag #

Patch Set 4 : Split out pattern and results in match routine #

Unified diffs Side-by-side diffs Delta from patch set Stats (+174 lines, -53 lines) Patch
M src/core/SkPictureFlat.h View 1 2 3 1 chunk +3 lines, -0 lines 0 comments Download
M src/core/SkPictureRecord.cpp View 1 2 3 6 chunks +171 lines, -53 lines 0 comments Download

Messages

Total messages: 5
robertphillips
11 years, 4 months ago (2013-03-05 17:04:16 UTC) #1
robertphillips
Comments on this?
11 years, 4 months ago (2013-03-07 18:07:54 UTC) #2
robertphillips
ping
11 years, 4 months ago (2013-03-11 16:23:32 UTC) #3
reed1
lgtm -- unless... does it make sense to separate out the input (request-op/*) and output ...
11 years, 4 months ago (2013-03-11 16:37:07 UTC) #4
robertphillips
11 years, 4 months ago (2013-03-12 15:40:03 UTC) #5
Done.

committed as r8106
Sign in to reply to this message.

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