Keyboard Shortcuts

File
u :up to issue
m :publish + mail comments
M :edit review message
j / k :jump to file after / before current file
J / K :jump to next file with a comment after / before current file
Side-by-side diff
i :toggle intra-line diffs
e :expand all comments
c :collapse all comments
s :toggle showing all comments
n / p :next / previous diff chunk or comment
N / P :next / previous comment
<Up> / <Down> :next / previous line
<Enter> :respond to / edit current comment
d :mark current comment as done
Issue
u :up to list of issues
m :publish + mail comments
j / k :jump to patch after / before current patch
o / <Enter> :open current patch in side-by-side view
i :open current patch in unified diff view
Issue List
j / k :jump to issue after / before current issue
o / <Enter> :open current issue
# : close issue
Comment/message editing
<Ctrl> + s or <Ctrl> + Enter :save comment
<Esc> :cancel edit
Rietveld Code Review Tool
Help | Bug tracker | Discussion group | Source code | Sign in
(791)
Issues Repositories Search
Open Issues | Closed Issues | All Issues | Sign in with your Google Account to create issues and add comments

Issue 6197072: optimizations for small drawRect calls

Can't Edit
Can't Publish+Mail
Start Review
Created:
13 years, 8 months ago by reed1
Modified:
13 years, 8 months ago
Reviewers:
TomH
CC:
skia-review_googlegroups.com, caryclark1, bsalomon
Base URL:
http://skia.googlecode.com/svn/trunk/
Visibility:
Public.
optimizations for small drawRect calls

Patch Set 1 #

Patch Set 2 : #

Patch Set 3 : #

Patch Set 4 : #

Patch Set 5 : #

Patch Set 6 : #

Created: 13 years, 8 months ago
Download [raw] [tar.bz2]
Unified diffs Side-by-side diffs Delta from patch set Stats (+37 lines, -26 lines) Patch
M src/core/SkCanvas.cpp View 1 2 3 4 5 3 chunks +37 lines, -26 lines 0 comments Download
Total messages: 6
|
reed1
Likely I will commit these piece-meal, but here is the sum of my experiments so ...
13 years, 8 months ago (2012年05月10日 20:25:00 UTC) #1
Likely I will commit these piece-meal, but here is the sum of my experiments so
far to speed up small drawRects (simulating dashing).
Bench runs are very noisy :( but I'm seeing ~25% faster on dash_4_rect
benchmark.
Will run some timings on linux (64bit) before I commit anything
Want to get some aggregate feeling of perf change on all benches (but how?)
Sign in to reply to this message.
reed1
~ 20% faster on linux (when I ran with -03 for both runs)
13 years, 8 months ago (2012年05月10日 20:50:31 UTC) #2
~ 20% faster on linux (when I ran with -03 for both runs)
Sign in to reply to this message.
TomH
The templates add a *lot* of code complexity to BlitRow_D32. Do they really yield that ...
13 years, 8 months ago (2012年05月10日 20:53:50 UTC) #3
The templates add a *lot* of code complexity to BlitRow_D32. Do they really
yield that much more performance than a naive unrolling? The only thing I see
you're winning is that the odd-man-out part of the loop (1..3) is unrolled.
For noisy timings, what -repeat count are you using?
My rule of thumb is -repeat 50 for 5% noise, -repeat 150 for 1%.
Sign in to reply to this message.
reed1
Good question about the value of the 1-3 tail. I will try expanding the test ...
13 years, 8 months ago (2012年05月10日 20:57:12 UTC) #4
Good question about the value of the 1-3 tail. I will try expanding the test to
time w/ and w/o those being templated.
Sign in to reply to this message.
reed1
removed template for catching the tail n%4 counts
13 years, 8 months ago (2012年05月11日 12:33:50 UTC) #5
removed template for catching the tail n%4 counts
Sign in to reply to this message.
reed1
committing the simpler pieces separately, to isolate these if there are any problems found.
13 years, 8 months ago (2012年05月11日 14:40:28 UTC) #6
committing the simpler pieces separately, to isolate these if there are any
problems found.
Sign in to reply to this message.
|
Powered by Google App Engine
RSS Feeds Recent Issues | This issue
This is Rietveld f62528b

AltStyle によって変換されたページ (->オリジナル) /