Start to vectorize SkTileGrid.

Start to vectorize SkTileGrid.

This adds Sk4x.h to help.

BUG=skia:3041
Committed: https://skia.googlesource.com/skia/+/90c7992bfc6330f070f7704d63372a0ec8410170

CQ_EXTRA_TRYBOTS=client.skia:Test-Ubuntu12-ShuttleA-GTX660-x86-Debug-Trybot

Committed: https://skia.googlesource.com/skia/+/958e9628d5f9a81aeafa78572cb4afc4b19a455a

[mtklein, 2014-10-10 15:45:27]

The CQ bit was checked by mtklein@google.com

[commit-bot: I haz the power, 2014-10-10 15:46:33]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/634543004/200001

[commit-bot: I haz the power, 2014-10-10 15:46:34]

Note for Reviewers:
The CQ is waiting for an approval.
If you believe that the CL is not ready yet, or if you would like to L-G-T-M
with comments then please uncheck the CQ checkbox.

Waiting for LGTM from valid reviewer(s) till 2014-10-10 21:46 UTC

[commit-bot: I haz the power, 2014-10-10 15:49:53]

The CQ bit was unchecked by commit-bot@chromium.org

[commit-bot: I haz the power, 2014-10-10 15:49:54]

Try jobs failed on following builders:
  Build-Ubuntu13.10-Clang-x86_64-Debug-Trybot on tryserver.skia
(http://108.170.220.76:10117/builders/Build-Ubuntu13.10-Clang-x86_64-Debug-Try...)
  Build-Ubuntu13.10-GCC4.8-Arm7-Debug-Android-Trybot on tryserver.skia
(http://108.170.220.76:10117/builders/Build-Ubuntu13.10-GCC4.8-Arm7-Debug-Andr...)

[mtklein, 2014-10-10 16:02:17]

The CQ bit was checked by mtklein@google.com

[commit-bot: I haz the power, 2014-10-10 16:02:49]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/634543004/310001

[commit-bot: I haz the power, 2014-10-10 16:02:50]

Note for Reviewers:
The CQ is waiting for an approval.
If you believe that the CL is not ready yet, or if you would like to L-G-T-M
with comments then please uncheck the CQ checkbox.

Waiting for LGTM from valid reviewer(s) till 2014-10-10 22:02 UTC

[reed1, 2014-10-10 16:06:10]

reed@google.com changed reviewers:
+ reed@google.com

[reed1, 2014-10-10 16:06:11]

https://codereview.chromium.org/634543004/diff/310001/src/core/Sk4x.h
File src/core/Sk4x.h (right):

https://codereview.chromium.org/634543004/diff/310001/src/core/Sk4x.h#newcode80
src/core/Sk4x.h:80: inline bool Sk4All(Sk4i v) {
as we start to actually use these, we might discuss the best patterns for naming
some of these (for clarity). e.g. this name wasn't obvious to me. perhaps some
convention for all predicate helpers? (dunno) Sk4_AND Sk4_OR etc. ? No real
suggestion at this time.

[commit-bot: I haz the power, 2014-10-10 16:06:18]

The CQ bit was unchecked by commit-bot@chromium.org

[commit-bot: I haz the power, 2014-10-10 16:06:19]

Try jobs failed on following builders:
  Test-Ubuntu13.10-GCE-NoGPU-x86_64-Release-Shared-Trybot on tryserver.skia
(http://108.170.220.120:10117/builders/Test-Ubuntu13.10-GCE-NoGPU-x86_64-Relea...)

[mtklein, 2014-10-10 17:21:49]

The CQ bit was checked by mtklein@google.com

[commit-bot: I haz the power, 2014-10-10 17:22:44]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/634543004/450001

[commit-bot: I haz the power, 2014-10-10 17:22:45]

Note for Reviewers:
The CQ is waiting for an approval.
If you believe that the CL is not ready yet, or if you would like to L-G-T-M
with comments then please uncheck the CQ checkbox.

Waiting for LGTM from valid reviewer(s) till 2014-10-10 23:22 UTC

[commit-bot: I haz the power, 2014-10-10 17:25:45]

The CQ bit was unchecked by commit-bot@chromium.org

[commit-bot: I haz the power, 2014-10-10 17:25:46]

Try jobs failed on following builders:
  Build-Ubuntu13.10-Clang-x86_64-Debug-Trybot on tryserver.skia
(http://108.170.220.76:10117/builders/Build-Ubuntu13.10-Clang-x86_64-Debug-Try...)

[mtklein, 2014-10-10 19:24:53]

The CQ bit was checked by mtklein@google.com

[commit-bot: I haz the power, 2014-10-10 19:25:38]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/634543004/600001

[commit-bot: I haz the power, 2014-10-10 19:25:39]

Note for Reviewers:
The CQ is waiting for an approval.
If you believe that the CL is not ready yet, or if you would like to L-G-T-M
with comments then please uncheck the CQ checkbox.

Waiting for LGTM from valid reviewer(s) till 2014-10-11 01:25 UTC

[commit-bot: I haz the power, 2014-10-10 19:28:48]

The CQ bit was unchecked by commit-bot@chromium.org

[commit-bot: I haz the power, 2014-10-10 19:28:49]

Try jobs failed on following builders:
  Build-Ubuntu13.10-GCC4.8-Arm7-Debug-Android-Trybot on tryserver.skia
(http://108.170.220.76:10117/builders/Build-Ubuntu13.10-GCC4.8-Arm7-Debug-Andr...)

[mtklein, 2014-10-10 19:59:13]

The CQ bit was checked by mtklein@google.com

[commit-bot: I haz the power, 2014-10-10 19:59:44]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/634543004/690001

[commit-bot: I haz the power, 2014-10-10 19:59:45]

Note for Reviewers:
The CQ is waiting for an approval.
If you believe that the CL is not ready yet, or if you would like to L-G-T-M
with comments then please uncheck the CQ checkbox.

Waiting for LGTM from valid reviewer(s) till 2014-10-11 01:59 UTC

[mtklein, 2014-10-10 20:24:02]

The CQ bit was checked by mtklein@google.com

[commit-bot: I haz the power, 2014-10-10 20:24:55]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/634543004/750001

[commit-bot: I haz the power, 2014-10-10 20:24:56]

Note for Reviewers:
The CQ is waiting for an approval.
If you believe that the CL is not ready yet, or if you would like to L-G-T-M
with comments then please uncheck the CQ checkbox.

Waiting for LGTM from valid reviewer(s) till 2014-10-11 02:24 UTC

[commit-bot: I haz the power, 2014-10-10 20:42:32]

The CQ bit was unchecked by commit-bot@chromium.org

[commit-bot: I haz the power, 2014-10-10 20:42:35]

Try jobs failed on following builders:
  Build-Win-VS2013-x86-Debug-Trybot on tryserver.skia
(http://108.170.220.76:10117/builders/Build-Win-VS2013-x86-Debug-Trybot/builds...)

[mtklein, 2014-10-10 21:07:51]

The CQ bit was checked by mtklein@google.com

[commit-bot: I haz the power, 2014-10-10 21:08:13]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/634543004/870001

[commit-bot: I haz the power, 2014-10-10 21:08:15]

Note for Reviewers:
The CQ is waiting for an approval.
If you believe that the CL is not ready yet, or if you would like to L-G-T-M
with comments then please uncheck the CQ checkbox.

Waiting for LGTM from valid reviewer(s) till 2014-10-11 03:08 UTC

[mtklein, 2014-10-10 21:18:19]

The CQ bit was checked by mtklein@google.com

[commit-bot: I haz the power, 2014-10-10 21:18:32]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/634543004/950001

[commit-bot: I haz the power, 2014-10-10 21:18:33]

Note for Reviewers:
The CQ is waiting for an approval.
If you believe that the CL is not ready yet, or if you would like to L-G-T-M
with comments then please uncheck the CQ checkbox.

Waiting for LGTM from valid reviewer(s) till 2014-10-11 03:18 UTC

[mtklein, 2014-10-10 21:26:07]

https://codereview.chromium.org/634543004/diff/310001/src/core/Sk4x.h
File src/core/Sk4x.h (right):

https://codereview.chromium.org/634543004/diff/310001/src/core/Sk4x.h#newcode80
src/core/Sk4x.h:80: inline bool Sk4All(Sk4i v) {
On 2014/10/10 16:06:11, reed1 wrote:
> as we start to actually use these, we might discuss the best patterns for
naming
> some of these (for clarity). e.g. this name wasn't obvious to me. perhaps some
> convention for all predicate helpers? (dunno) Sk4_AND Sk4_OR etc. ? No real
> suggestion at this time.

See what you think of the API at PS 20?

[mtklein, 2014-10-10 23:53:24]

The CQ bit was checked by mtklein@google.com

[commit-bot: I haz the power, 2014-10-10 23:53:52]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/634543004/1030001

[commit-bot: I haz the power, 2014-10-10 23:53:54]

Note for Reviewers:
The CQ is waiting for an approval.
If you believe that the CL is not ready yet, or if you would like to L-G-T-M
with comments then please uncheck the CQ checkbox.

Waiting for LGTM from valid reviewer(s) till 2014-10-11 05:53 UTC

[mtklein, 2014-10-15 20:30:44]

The CQ bit was checked by mtklein@google.com

[commit-bot: I haz the power, 2014-10-15 20:31:44]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/634543004/1300001

[commit-bot: I haz the power, 2014-10-15 20:31:46]

Note for Reviewers:
The CQ is waiting for an approval.
If you believe that the CL is not ready yet, or if you would like to L-G-T-M
with comments then please uncheck the CQ checkbox.

Waiting for LGTM from valid reviewer(s) till 2014-10-16 02:31 UTC

[reed1, 2014-10-15 20:48:12]

https://codereview.chromium.org/634543004/diff/1300001/src/core/SkTileGrid.cpp
File src/core/SkTileGrid.cpp (right):

https://codereview.chromium.org/634543004/diff/1300001/src/core/SkTileGrid.cp...
src/core/SkTileGrid.cpp:135: const int tilesHit = (grid.fRight - grid.fLeft + 1)
* (grid.fBottom - grid.fTop + 1);
I see a +1 here, and a change from < to <= down below.

Not trying to pick on this CL per-se, but I see this change as well as just
scalers -> 4f.

1. I'm sure I don't understand either of the two (i.e. I don't understand the
trickiness of this logic either way)?
2. Is this change an integral part of the Sk4x change?

I'm kind of ok either way, but I do think that switching to Sk4x makes the code
faster, but definitely harder to understand.

[mtklein, 2014-10-15 20:55:28]

https://codereview.chromium.org/634543004/diff/1300001/src/core/SkTileGrid.cpp
File src/core/SkTileGrid.cpp (right):

https://codereview.chromium.org/634543004/diff/1300001/src/core/SkTileGrid.cp...
src/core/SkTileGrid.cpp:135: const int tilesHit = (grid.fRight - grid.fLeft + 1)
* (grid.fBottom - grid.fTop + 1);
On 2014/10/15 20:48:11, reed1 wrote:
> I see a +1 here, and a change from < to <= down below.
> 
> Not trying to pick on this CL per-se, but I see this change as well as just
> scalers -> 4f.
> 
> 1. I'm sure I don't understand either of the two (i.e. I don't understand the
> trickiness of this logic either way)?
> 2. Is this change an integral part of the Sk4x change?
> 
> I'm kind of ok either way, but I do think that switching to Sk4x makes the
code
> faster, but definitely harder to understand.

Ah, yeah, you're watching work in progress.  That +1/<= change is going to spin
off into a prequel CL.

[mtklein, 2014-10-15 21:47:22]

Have a look now Mike?

This is a delta off head + https://codereview.chromium.org/658913002/.

[mtklein, 2014-10-16 17:18:39]

The CQ bit was checked by mtklein@google.com

[commit-bot: I haz the power, 2014-10-16 17:19:47]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/634543004/1410001

[commit-bot: I haz the power, 2014-10-16 17:19:49]

Note for Reviewers:
The CQ is waiting for an approval.
If you believe that the CL is not ready yet, or if you would like to L-G-T-M
with comments then please uncheck the CQ checkbox.

Waiting for LGTM from valid reviewer(s) till 2014-10-16 23:19 UTC

[mtklein, 2014-10-16 18:42:33]

A comparison run on Nexus 5 (GCC but no SIMD) shows this should be a
in-the-noise change:
                  desk_baidu.skp	 462us ->  526us	1.1x
     desk_jsfiddlehumperclip.skp	 187us ->  211us	1.1x
                   desk_ebay.skp	1.15ms -> 1.27ms	1.1x
           tabl_worldjournal.skp	2.42ms -> 2.62ms	1.1x
               tabl_slashdot.skp	 575us ->  622us	1.1x
             desk_googleplus.skp	2.22ms -> 2.38ms	1.1x
           desk_yahooanswers.skp	 651us ->  698us	1.1x
         desk_jsfiddlebigcar.skp	 157us ->  167us	1.1x
             desk_chalkboard.skp	 2.5ms -> 2.65ms	1.1x
               desk_samoasvg.skp	1.75ms -> 1.85ms	1.1x
                tabl_nytimes.skp	1.28ms -> 1.35ms	1.1x
                  tabl_ukwsj.skp	 1.5ms -> 1.58ms	1.1x
    desk_rectangletransition.skp	48.5us ->   51us	1.1x
               tabl_engadget.skp	2.27ms -> 2.37ms	1x
                 desk_mapsvg.skp	5.12ms -> 5.34ms	1x
             desk_googlehome.skp	 178us ->  185us	1x
            desk_gmailthread.skp	1.01ms -> 1.05ms	1x
               desk_fontwipe.skp	32.6us -> 33.9us	1x
              desk_pinterest.skp	 200us ->  208us	1x
                   tabl_cnet.skp	2.41ms ->  2.5ms	1x
             desk_forecastio.skp	 572us ->  593us	1x
              tabl_vnexpress.skp	1.47ms -> 1.53ms	1x
             tabl_deviantart.skp	 761us ->  787us	1x
                  tabl_gspro.skp	 333us ->  344us	1x
                 desk_carsvg.skp	1.75ms ->  1.8ms	1x
            desk_pokemonwiki.skp	32.4ms -> 33.3ms	1x
                desk_twitter.skp	1.58ms -> 1.62ms	1x
desk_googlespreadsheetdashed.skp	8.85ms -> 9.05ms	1x
                 tabl_nofolo.skp	 275us ->  280us	1x
              mobi_wikipedia.skp	4.56ms -> 4.65ms	1x
            desk_oldinboxapp.skp	24.6us -> 25.1us	1x
                   tabl_hsfi.skp	2.02ms -> 2.05ms	1x
                desk_wowwiki.skp	7.33ms -> 7.45ms	1x
                tabl_sahadan.skp	 335us ->  340us	1x
                  tabl_gmail.skp	 143us ->  145us	1x
         tabl_googlecalendar.skp	 753us ->  763us	1x
                desk_weather.skp	 822us ->  831us	1x
           tabl_cuteoverload.skp	3.41ms -> 3.45ms	1x
          tabl_androidpolice.skp	4.11ms ->  4.1ms	1x
                 desk_sfgate.skp	1.48ms -> 1.47ms	0.99x
      desk_googlespreadsheet.skp	1.37ms -> 1.34ms	0.98x
                tabl_mozilla.skp	17.2ms -> 16.9ms	0.98x
                desk_booking.skp	3.06ms ->    3ms	0.98x
               tabl_frantzen.skp	 467us ->  455us	0.97x
                 tabl_pravda.skp	1.12ms -> 1.09ms	0.97x
          desk_css3gradients.skp	1.31ms -> 1.28ms	0.97x
                desk_youtube.skp	1.76ms ->  1.7ms	0.97x
             desk_mobilenews.skp	1.95ms -> 1.88ms	0.96x
            tabl_gamedeksiam.skp	3.29ms -> 3.14ms	0.96x
      tabl_culturalsolutions.skp	1.72ms -> 1.64ms	0.96x
                   desk_espn.skp	1.21ms -> 1.15ms	0.96x
               desk_tigersvg.skp	 280us ->  266us	0.95x
                    tabl_mlb.skp	 879us ->  835us	0.95x
                 desk_amazon.skp	 610us ->  578us	0.95x
               tabl_techmeme.skp	 560us ->  530us	0.95x
               desk_facebook.skp	 931us ->  878us	0.94x
            tabl_mercurynews.skp	 1.1ms -> 1.02ms	0.93x
               desk_linkedin.skp	 850us ->  790us	0.93x
         desk_youtubetvvideo.skp	73.1us -> 67.8us	0.93x
                    desk_gws.skp	 704us ->  650us	0.92x
                desk_blogger.skp	1.46ms ->  1.3ms	0.89x

Going to try now explicitly forcing Sk4x_portable.h to simulate Windows.

[mtklein, 2014-10-16 19:24:10]

Added some microbeches and switched intersection back to SkRect::Intersect based
off their results:
 
57M	48	6.02µs	6.03µs	6.04µs	6.16µs	1%	▁▁▁█▁▁▁▁▁▁	nonrendering	geo_rect_Intersects_4f
  57M	80	3.55µs	3.55µs	3.58µs	3.8µs	2%	▁▁▁█▁▁▁▁▁▁	nonrendering	geo_rect_sort_4f
  57M	67	4.25µs	4.25µs	4.29µs	4.58µs	2%	▁▁▁▁▁▁█▁▁▁	nonrendering	geo_rect_sort
 
57M	92	2.85µs	2.85µs	2.85µs	2.85µs	0%	█▂▁▁▂▁▂▃▃▂	nonrendering	geo_rect_Intersects

I wrote the microbenchmarks as drop-in replacements for SkRect::sort() and
SkRect::Intersects().  sort()'s a little
faster with Sk4f, and Intersects is about 2x slower.  This is somewhere where
Sk4x_sse.h could help.

Because SkRect::Intersects was the winner there, I switched the SkTileGrid code
back to use it instead of the Sk4f code.
The results are neutral-to-positive on recording now (this is Clang generating
SSE2):
         desk_youtubetvvideo.skp	11.4us -> 12.1us	1.1x
          desk_css3gradients.skp	 200us ->  207us	1x
                   desk_ebay.skp	 194us ->  199us	1x
                 desk_carsvg.skp	 272us ->  277us	1x
            desk_oldinboxapp.skp	4.18us -> 4.25us	1x
               desk_fontwipe.skp	5.28us -> 5.34us	1x
      tabl_culturalsolutions.skp	 247us ->  249us	1x
                   tabl_cnet.skp	 369us ->  372us	1x
               desk_tigersvg.skp	41.8us -> 42.1us	1x
             desk_googlehome.skp	31.9us -> 31.7us	0.99x
                desk_twitter.skp	 245us ->  242us	0.99x
         desk_jsfiddlebigcar.skp	29.9us -> 29.6us	0.99x
               tabl_frantzen.skp	73.9us -> 72.9us	0.99x
              mobi_wikipedia.skp	 565us ->  555us	0.98x
                   desk_espn.skp	 190us ->  186us	0.98x
            desk_gmailthread.skp	 172us ->  168us	0.98x
                tabl_nytimes.skp	 196us ->  192us	0.98x
            tabl_mercurynews.skp	 169us ->  166us	0.98x
             desk_chalkboard.skp	 306us ->  300us	0.98x
                   tabl_hsfi.skp	 311us ->  304us	0.98x
            tabl_gamedeksiam.skp	 481us ->  470us	0.98x
             tabl_deviantart.skp	 116us ->  114us	0.98x
           tabl_cuteoverload.skp	 508us ->  496us	0.98x
                desk_wowwiki.skp	1.01ms ->  986us	0.98x
                desk_blogger.skp	 210us ->  205us	0.98x
          tabl_androidpolice.skp	 661us ->  645us	0.98x
                  tabl_ukwsj.skp	 250us ->  244us	0.98x
desk_googlespreadsheetdashed.skp	1.03ms ->    1ms	0.97x
                 desk_mapsvg.skp	 570us ->  554us	0.97x
             tabl_googleblog.skp	 317us ->  308us	0.97x
                 tabl_pravda.skp	 162us ->  157us	0.97x
             desk_mobilenews.skp	 275us ->  267us	0.97x
               tabl_slashdot.skp	 102us -> 98.6us	0.97x
               desk_linkedin.skp	 135us ->  130us	0.97x
         tabl_googlecalendar.skp	 137us ->  132us	0.97x
                tabl_mozilla.skp	2.16ms -> 2.09ms	0.97x
     desk_jsfiddlehumperclip.skp	  33us -> 31.9us	0.97x
                  desk_baidu.skp	83.2us -> 80.4us	0.97x
                  tabl_gspro.skp	55.6us -> 53.7us	0.97x
               tabl_engadget.skp	 388us ->  375us	0.97x
                desk_youtube.skp	 289us ->  278us	0.96x
               desk_facebook.skp	 148us ->  142us	0.96x
                 tabl_nofolo.skp	45.8us ->   44us	0.96x
                    desk_gws.skp	 113us ->  109us	0.96x
           tabl_worldjournal.skp	 401us ->  385us	0.96x
             desk_forecastio.skp	98.9us -> 94.2us	0.95x
             desk_googleplus.skp	 370us ->  349us	0.94x
                tabl_sahadan.skp	57.5us -> 54.2us	0.94x
              desk_pinterest.skp	  34us ->   32us	0.94x
    desk_rectangletransition.skp	8.83us -> 8.28us	0.94x
                desk_weather.skp	 154us ->  144us	0.94x
                 desk_sfgate.skp	 212us ->  198us	0.93x
                 desk_amazon.skp	 103us -> 95.9us	0.93x
                    tabl_mlb.skp	 152us ->  140us	0.92x
               tabl_techmeme.skp	  93us -> 84.7us	0.91x

[mtklein, 2014-10-16 19:24:13]

Added some microbeches and switched intersection back to SkRect::Intersect based
off their results:
 
57M	48	6.02µs	6.03µs	6.04µs	6.16µs	1%	▁▁▁█▁▁▁▁▁▁	nonrendering	geo_rect_Intersects_4f
  57M	80	3.55µs	3.55µs	3.58µs	3.8µs	2%	▁▁▁█▁▁▁▁▁▁	nonrendering	geo_rect_sort_4f
  57M	67	4.25µs	4.25µs	4.29µs	4.58µs	2%	▁▁▁▁▁▁█▁▁▁	nonrendering	geo_rect_sort
 
57M	92	2.85µs	2.85µs	2.85µs	2.85µs	0%	█▂▁▁▂▁▂▃▃▂	nonrendering	geo_rect_Intersects

I wrote the microbenchmarks as drop-in replacements for SkRect::sort() and
SkRect::Intersects().  sort()'s a little
faster with Sk4f, and Intersects is about 2x slower.  This is somewhere where
Sk4x_sse.h could help.

Because SkRect::Intersects was the winner there, I switched the SkTileGrid code
back to use it instead of the Sk4f code.
The results are neutral-to-positive on recording now (this is Clang generating
SSE2):
         desk_youtubetvvideo.skp	11.4us -> 12.1us	1.1x
          desk_css3gradients.skp	 200us ->  207us	1x
                   desk_ebay.skp	 194us ->  199us	1x
                 desk_carsvg.skp	 272us ->  277us	1x
            desk_oldinboxapp.skp	4.18us -> 4.25us	1x
               desk_fontwipe.skp	5.28us -> 5.34us	1x
      tabl_culturalsolutions.skp	 247us ->  249us	1x
                   tabl_cnet.skp	 369us ->  372us	1x
               desk_tigersvg.skp	41.8us -> 42.1us	1x
             desk_googlehome.skp	31.9us -> 31.7us	0.99x
                desk_twitter.skp	 245us ->  242us	0.99x
         desk_jsfiddlebigcar.skp	29.9us -> 29.6us	0.99x
               tabl_frantzen.skp	73.9us -> 72.9us	0.99x
              mobi_wikipedia.skp	 565us ->  555us	0.98x
                   desk_espn.skp	 190us ->  186us	0.98x
            desk_gmailthread.skp	 172us ->  168us	0.98x
                tabl_nytimes.skp	 196us ->  192us	0.98x
            tabl_mercurynews.skp	 169us ->  166us	0.98x
             desk_chalkboard.skp	 306us ->  300us	0.98x
                   tabl_hsfi.skp	 311us ->  304us	0.98x
            tabl_gamedeksiam.skp	 481us ->  470us	0.98x
             tabl_deviantart.skp	 116us ->  114us	0.98x
           tabl_cuteoverload.skp	 508us ->  496us	0.98x
                desk_wowwiki.skp	1.01ms ->  986us	0.98x
                desk_blogger.skp	 210us ->  205us	0.98x
          tabl_androidpolice.skp	 661us ->  645us	0.98x
                  tabl_ukwsj.skp	 250us ->  244us	0.98x
desk_googlespreadsheetdashed.skp	1.03ms ->    1ms	0.97x
                 desk_mapsvg.skp	 570us ->  554us	0.97x
             tabl_googleblog.skp	 317us ->  308us	0.97x
                 tabl_pravda.skp	 162us ->  157us	0.97x
             desk_mobilenews.skp	 275us ->  267us	0.97x
               tabl_slashdot.skp	 102us -> 98.6us	0.97x
               desk_linkedin.skp	 135us ->  130us	0.97x
         tabl_googlecalendar.skp	 137us ->  132us	0.97x
                tabl_mozilla.skp	2.16ms -> 2.09ms	0.97x
     desk_jsfiddlehumperclip.skp	  33us -> 31.9us	0.97x
                  desk_baidu.skp	83.2us -> 80.4us	0.97x
                  tabl_gspro.skp	55.6us -> 53.7us	0.97x
               tabl_engadget.skp	 388us ->  375us	0.97x
                desk_youtube.skp	 289us ->  278us	0.96x
               desk_facebook.skp	 148us ->  142us	0.96x
                 tabl_nofolo.skp	45.8us ->   44us	0.96x
                    desk_gws.skp	 113us ->  109us	0.96x
           tabl_worldjournal.skp	 401us ->  385us	0.96x
             desk_forecastio.skp	98.9us -> 94.2us	0.95x
             desk_googleplus.skp	 370us ->  349us	0.94x
                tabl_sahadan.skp	57.5us -> 54.2us	0.94x
              desk_pinterest.skp	  34us ->   32us	0.94x
    desk_rectangletransition.skp	8.83us -> 8.28us	0.94x
                desk_weather.skp	 154us ->  144us	0.94x
                 desk_sfgate.skp	 212us ->  198us	0.93x
                 desk_amazon.skp	 103us -> 95.9us	0.93x
                    tabl_mlb.skp	 152us ->  140us	0.92x
               tabl_techmeme.skp	  93us -> 84.7us	0.91x

[mtklein, 2014-10-16 20:39:17]

The CQ bit was checked by mtklein@google.com

[commit-bot: I haz the power, 2014-10-16 20:40:21]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/634543004/1510001

[commit-bot: I haz the power, 2014-10-16 20:40:23]

Note for Reviewers:
The CQ is waiting for an approval.
If you believe that the CL is not ready yet, or if you would like to L-G-T-M
with comments then please uncheck the CQ checkbox.

Waiting for LGTM from valid reviewer(s) till 2014-10-17 02:40 UTC

[mtklein, 2014-10-16 20:50:55]

The CQ bit was checked by mtklein@google.com

[commit-bot: I haz the power, 2014-10-16 20:51:25]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/634543004/1530001

[commit-bot: I haz the power, 2014-10-16 20:51:27]

Note for Reviewers:
The CQ is waiting for an approval.
If you believe that the CL is not ready yet, or if you would like to L-G-T-M
with comments then please uncheck the CQ checkbox.

Waiting for LGTM from valid reviewer(s) till 2014-10-17 02:51 UTC

[commit-bot: I haz the power, 2014-10-16 20:56:08]

The CQ bit was unchecked by commit-bot@chromium.org

[commit-bot: I haz the power, 2014-10-16 20:56:10]

Try jobs failed on following builders:
  Build-Win-VS2013-x86-Debug-Trybot on client.skia.compile
(http://build.chromium.org/p/client.skia.compile/builders/Build-Win-VS2013-x86...)

[mtklein, 2014-10-16 20:59:13]

The CQ bit was checked by mtklein@google.com

[commit-bot: I haz the power, 2014-10-16 20:59:43]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/634543004/1550001

[commit-bot: I haz the power, 2014-10-16 20:59:45]

Note for Reviewers:
The CQ is waiting for an approval.
If you believe that the CL is not ready yet, or if you would like to L-G-T-M
with comments then please uncheck the CQ checkbox.

Waiting for LGTM from valid reviewer(s) till 2014-10-17 02:59 UTC

[commit-bot: I haz the power, 2014-10-16 21:04:59]

The CQ bit was unchecked by commit-bot@chromium.org

[commit-bot: I haz the power, 2014-10-16 21:05:01]

Try jobs failed on following builders:
  Build-Win-VS2013-x86-Debug-Trybot on client.skia.compile
(http://build.chromium.org/p/client.skia.compile/builders/Build-Win-VS2013-x86...)

[mtklein, 2014-10-16 21:27:08]

The CQ bit was checked by mtklein@google.com

[commit-bot: I haz the power, 2014-10-16 21:27:50]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/634543004/1590001

[commit-bot: I haz the power, 2014-10-16 21:27:51]

Note for Reviewers:
The CQ is waiting for an approval.
If you believe that the CL is not ready yet, or if you would like to L-G-T-M
with comments then please uncheck the CQ checkbox.

Waiting for LGTM from valid reviewer(s) till 2014-10-17 03:27 UTC

[commit-bot: I haz the power, 2014-10-16 21:28:05]

The CQ bit was unchecked by commit-bot@chromium.org

[commit-bot: I haz the power, 2014-10-16 21:28:07]

Presubmit check for 634543004-1590001 failed and returned exit status 1.

Running presubmit commit checks ...

** Presubmit ERRORS **
Since the CL is editing public API, you must have an LGTM from one of:
('reed@chromium.org', 'reed@google.com', 'bsalomon@chromium.org',
'bsalomon@google.com', 'djsollen@chromium.org', 'djsollen@google.com')

Presubmit checks took 1.3s to calculate.

Was the presubmit check useful? If not, run "git cl presubmit -v"
to figure out which PRESUBMIT.py was run, then run git blame
on the file to figure out who to ask for help.

[mtklein, 2014-10-16 21:30:38]

The CQ bit was checked by mtklein@google.com

[commit-bot: I haz the power, 2014-10-16 21:30:50]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/634543004/1590001

[reed1, 2014-10-16 21:46:25]

lgtm

[mtklein, 2014-10-16 22:13:20]

The CQ bit was checked by mtklein@google.com

[commit-bot: I haz the power, 2014-10-16 22:14:06]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/634543004/1430002

[commit-bot: I haz the power, 2014-10-16 22:21:48]

Committed patchset #45 (id:1430002) as 90c7992bfc6330f070f7704d63372a0ec8410170

[mtklein, 2014-10-16 22:26:23]

A revert of this CL (patchset #45 id:1430002) has been created in
https://codereview.chromium.org/663663002/ by mtklein@google.com.

The reason for reverting is: Many GCC bots missing __builtin_shuffle, e.g.
Test-Ubuntu12-ShuttleA-GTX660-x86-Debug-Trybot..

[mtklein, 2014-10-20 13:48:08]

The CQ bit was checked by mtklein@google.com

[commit-bot: I haz the power, 2014-10-20 13:48:43]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/634543004/1670001

[commit-bot: I haz the power, 2014-10-20 14:12:40]

Committed patchset #48 (id:1670001) as 958e9628d5f9a81aeafa78572cb4afc4b19a455a

[mtklein, 2014-10-20 17:42:24]

A revert of this CL (patchset #48 id:1670001) has been created in
https://codereview.chromium.org/637863005/ by mtklein@google.com.

The reason for reverting is: breaks chrome GPU debug bots.