AuroraMiddleware/skia2
1
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity
12,567 Commits 1 Branch 0 Tags 544 MiB
baa860c297
Commit Graph

14 Commits

Author SHA1 Message Date
mtklein
90c471e73f Refine bench_record and bench_playback: 
- use high-precision wall timer only
  - warm caches once before measuring
  - measure independent samples, calculating statistics
  - add --verbose to control how much data we output

Also removed some unloved features from bench_record.

BUG=skia:
R=jcgregorio@google.com, mtklein@google.com

Author: mtklein@chromium.org

Review URL: https://codereview.chromium.org/338203002
 2014-06-16 14:04:34 -07:00
robertphillips
9f1c241e0d Remove SkPicture::kUsePathBoundsForClip_RecordingFlag 
The real question is whether we ever want to record a picture without using the path bounds for a conservative (but faster) clip answer?

R=reed@google.com, mtklein@google.com, djsollen@google.com, scroggo@google.com

Author: robertphillips@google.com

Review URL: https://codereview.chromium.org/316143003
 2014-06-09 06:25:34 -07:00
commit-bot@chromium.org
a095041f51 Remove SkRecorder's kWriteOnly mode. 
I'm soon going to have SkRecorder start calling getTotalMatrix(), which
would be broken in write-only mode.  That change is big and nebulous,
but it's clear kWriteOnly needs to go, so we might as well kill it now.

My notes in bench_playback about kWriteOnly mode being important were
probably overly cautious.  I now think this is a fair enough comparison
even re-recording into a read-write canvas.

BUG=skia:2378
R=fmalita@chromium.org, mtklein@google.com

Author: mtklein@chromium.org

Review URL: https://codereview.chromium.org/290653004

git-svn-id: http://skia.googlecode.com/svn/trunk@14963 2bbb7eff-a529-9590-31e7-b0007b416f81
 2014-05-29 16:52:40 +00:00
commit-bot@chromium.org
8fe89d3124 Use a tilegrid for bench_playback. 
Using --tile wasn't a fair comparison: SkPicture didn't get to optimize its
playback, but SkRecord did.  Numbers here are starting to agree with what I saw
in telemetry: SkRecord with culling is on average at 0.8-0.9 of the speed of
SkPicture (10-30% slower) than SkPicture SkPicture with a tile grid, but some
in some cases it's down at 0.3, taking almost 4x as long to draw a 256x256 tile.

0.4x tabl_mercurynews.skp 0.109889 0.252483
0.5x tabl_cnet.skp 0.250367 0.527789
1.0x desk_mapsvg.skp 1.48698 1.49939
1.2x desk_yahoogames.skp 0.07745 0.065343
0.9x tabl_mlb.skp 0.294509 0.312505
0.8x tabl_hsfi.skp 0.250678 0.316337
0.6x tabl_worldjournal.skp 0.510204 0.828798
0.7x desk_facebook.skp 0.318981 0.483744
0.9x desk_googlehome.skp 0.06983 0.078964
0.9x desk_tigersvg.skp 0.019232 0.022611
1.0x desk_youtubetvvideo.skp 0.401614 0.401862
1.1x desk_amazon.skp 0.415552 0.362111
0.7x desk_yahooanswers.skp 0.237995 0.358659
0.7x desk_twitter.skp 0.323466 0.439772
1.2x desk_fontwipe.skp 0.121577 0.097845
1.1x desk_css3gradients.skp 0.395335 0.362087
0.9x desk_googlespreadsheet.skp 0.362056 0.406688
1.0x tabl_frantzen.skp 15.2475 14.8521
0.6x desk_forecastio.skp 0.154236 0.276881
0.7x desk_pinterest.skp 0.077433 0.11324
0.5x tabl_cuteoverload.skp 0.173014 0.361815
0.5x tabl_androidpolice.skp 0.225903 0.465608
0.6x tabl_vnexpress.skp 0.151075 0.241104
0.6x tabl_nytimes.skp 0.06515 0.117229
0.8x desk_ebay.skp 0.353437 0.422586
0.8x tabl_deviantart.skp 0.223103 0.295496
1.1x tabl_culturalsolutions.skp 0.551348 0.506104
0.9x mobi_wikipedia.skp 0.302603 0.337292
0.6x desk_linkedin.skp 0.165102 0.276149
0.6x tabl_digg.skp 0.159074 0.281529
1.2x desk_oldinboxapp.skp 0.019815 0.017146
1.0x tabl_nofolo.skp 5.18914 5.42784
0.6x desk_jsfiddlehumperclip.skp 0.076869 0.133665
1.0x desk_blogger.skp 0.206702 0.19688
0.8x desk_espn.skp 0.346386 0.451871
1.0x desk_samoasvg.skp 3.60037 3.43147
0.8x desk_booking.skp 0.338818 0.438267
1.3x desk_yahoosports.skp 0.603811 0.470052
0.6x tabl_engadget.skp 0.248995 0.404372
0.7x desk_wowwiki.skp 0.325801 0.479559
0.8x tabl_ukwsj.skp 0.326985 0.405971
0.9x tabl_gmail.skp 0.2604 0.276744
0.9x desk_googleplus.skp 8.08874 8.71105
1.1x tabl_slashdot.skp 0.559332 0.524313
0.9x desk_gmailthread.skp 0.671569 0.74134
0.8x desk_weather.skp 0.194162 0.247681
1.3x desk_rectangletransition.skp 0.218372 0.173685
0.6x tabl_gamedeksiam.skp 0.243374 0.4134
0.9x desk_jsfiddlebigcar.skp 0.162282 0.186035
0.3x desk_pokemonwiki.skp 0.202878 0.778934
0.7x tabl_googleblog.skp 0.141203 0.209971
0.8x desk_sfgate.skp 0.229078 0.275248
1.0x desk_chalkboard.skp 2.28916 2.20101
0.9x tabl_googlecalendar.skp 0.687155 0.75413
1.1x tabl_gspro.skp 0.186399 0.169522
0.5x desk_googlespreadsheetdashed.skp 0.512622 1.02815
0.6x desk_mobilenews.skp 0.183506 0.293929
0.8x tabl_techmeme.skp 0.193074 0.229106
1.2x desk_carsvg.skp 5.33047 4.34058
0.5x desk_youtube.skp 0.132663 0.248236
0.7x tabl_pravda.skp 0.21644 0.308788
1.1x desk_gws.skp 0.698107 0.625371
0.5x tabl_mozilla.skp 0.189638 0.386801
1.0x desk_baidu.skp 0.353438 0.355965
1.1x tabl_sahadan.skp 0.476226 0.416737
summary: ≥0.3x (A 0.9x, G 0.8x) ≤1.3x

Whole SKP playback speed is fine of course.

BUG=skia:2378
R=fmalita@chromium.org, mtklein@google.com

Author: mtklein@chromium.org

Review URL: https://codereview.chromium.org/277653002

git-svn-id: http://skia.googlecode.com/svn/trunk@14673 2bbb7eff-a529-9590-31e7-b0007b416f81
 2014-05-09 15:00:10 +00:00
commit-bot@chromium.org
172eb1baaf Add --timescale to bench_record and bench_playback. 
Default bench_record to report us, bench_playback ms.

BUG=skia:2378
R=mtklein@google.com

Author: mtklein@chromium.org

Review URL: https://codereview.chromium.org/253843004

git-svn-id: http://skia.googlecode.com/svn/trunk@14417 2bbb7eff-a529-9590-31e7-b0007b416f81
 2014-04-28 19:41:17 +00:00
commit-bot@chromium.org
8400b23ce6 use BenchTimer, print in µs 
BUG=skia:2378
R=bungeman@google.com, mtklein@google.com

Author: mtklein@chromium.org

Review URL: https://codereview.chromium.org/257563006

git-svn-id: http://skia.googlecode.com/svn/trunk@14405 2bbb7eff-a529-9590-31e7-b0007b416f81
 2014-04-28 15:30:02 +00:00
commit-bot@chromium.org
d0306a1593 Revert of Add nanosecond timer. (https://codereview.chromium.org/250243002/) 
Reason for revert:
breaks EVERYTHING

Original issue's description:
> Add nanosecond timer.
>
> I've been finding it hard to get enough resolution out of our existing timers when measuring really tiny pictures.
>
> BUG=skia:2378
>
> Committed: http://code.google.com/p/skia/source/detail?r=14362

R=bsalomon@google.com, bungeman@google.com, mtklein@chromium.org
TBR=bsalomon@google.com, bungeman@google.com, mtklein@chromium.org
NOTREECHECKS=true
NOTRY=true
BUG=skia:2378

Author: mtklein@google.com

Review URL: https://codereview.chromium.org/258703002

git-svn-id: http://skia.googlecode.com/svn/trunk@14364 2bbb7eff-a529-9590-31e7-b0007b416f81
 2014-04-24 20:17:24 +00:00
commit-bot@chromium.org
74b43a9d4c Add nanosecond timer. 
I've been finding it hard to get enough resolution out of our existing timers when measuring really tiny pictures.

BUG=skia:2378
R=bsalomon@google.com, mtklein@google.com, bungeman@google.com

Author: mtklein@chromium.org

Review URL: https://codereview.chromium.org/250243002

git-svn-id: http://skia.googlecode.com/svn/trunk@14362 2bbb7eff-a529-9590-31e7-b0007b416f81
 2014-04-24 20:04:20 +00:00
commit-bot@chromium.org
ad8ce572f6 anticipate more optimizations by renaming some files and methods 
also, call the new SkRecordOptimize in bench_playback

BUG=skia:2378
R=fmalita@chromium.org, mtklein@google.com

Author: mtklein@chromium.org

Review URL: https://codereview.chromium.org/243243003

git-svn-id: http://skia.googlecode.com/svn/trunk@14277 2bbb7eff-a529-9590-31e7-b0007b416f81
 2014-04-21 15:03:36 +00:00
commit-bot@chromium.org
5da5b59764 Add --match to bench tools, extra decimal to bench_pictures. 
BUG=skia:2378
R=fmalita@chromium.org, mtklein@google.com

Author: mtklein@chromium.org

Review URL: https://codereview.chromium.org/243273002

git-svn-id: http://skia.googlecode.com/svn/trunk@14276 2bbb7eff-a529-9590-31e7-b0007b416f81
 2014-04-21 14:59:59 +00:00
commit-bot@chromium.org
b17a24fedb SkRecord: turn on cull annotation pass in public API. 
Also, switch Skia internal tools over to use the public headers where
possible.  Where it's not, the tools -Isrc/record explicitly now, and
if it's not obvious, note why they don't use SkRecording.h.

BUG=skia:2378
R=reed@google.com, mtklein@google.com, fmalita@chromium.org

Author: mtklein@chromium.org

Review URL: https://codereview.chromium.org/231853006

git-svn-id: http://skia.googlecode.com/svn/trunk@14191 2bbb7eff-a529-9590-31e7-b0007b416f81
 2014-04-14 20:33:05 +00:00
commit-bot@chromium.org
d6489c957f use the correct rowBytes so Debug build doesn't assert 
BUG=skia:2378
R=reed@google.com, mtklein@google.com

Author: mtklein@chromium.org

Review URL: https://codereview.chromium.org/235453002

git-svn-id: http://skia.googlecode.com/svn/trunk@14163 2bbb7eff-a529-9590-31e7-b0007b416f81
 2014-04-11 19:06:46 +00:00
commit-bot@chromium.org
4bffdf2a10 use SkIntToScalar, fix windows build 
NOTRY=true
NOTREECHECKS=true
BUG=skia:
R=mtklein@google.com

Author: mtklein@chromium.org

Review URL: https://codereview.chromium.org/234913004

git-svn-id: http://skia.googlecode.com/svn/trunk@14155 2bbb7eff-a529-9590-31e7-b0007b416f81
 2014-04-11 16:13:54 +00:00
commit-bot@chromium.org
ba73d28bef Add bench_playback, for quick profiling of SKR playback. 
It is not completely fair to compare --skr and --noskr numbers.  SKR looks
unfairly good because some optimizations are baked into our SKPs at record
time.  But, at least, by using a kWriteOnly_Mode SkRecorder, we prevent
SkPicturePlayback from compounding that unfairness.  SkRecordDraw must handle
its own playback-time optimizations (quickrejects) on its own.

This code should look suspiciously similar to bench_record.

BUG=skia:2378
R=fmalita@chromium.org, mtklein@google.com

Author: mtklein@chromium.org

Review URL: https://codereview.chromium.org/233833002

git-svn-id: http://skia.googlecode.com/svn/trunk@14151 2bbb7eff-a529-9590-31e7-b0007b416f81
 2014-04-11 15:53:39 +00:00