Graphic optimizations 1 #4708

p01 · 2014-04-29T15:26:04Z

The main optimizations here are in the CalGrayCS conversion and JPG decoder's getLinearizedBlockData and convertCmykToRgb which got 2.16x and 1.5x faster respectively while using less memory ( and generating less garbage collection ).

Snuffleupagus · 2014-04-29T15:45:52Z

/botio test

pdfjsbot · 2014-04-29T15:45:54Z

From: Bot.io (Windows)

Received

Command cmd_test from @Snuffleupagus received. Current queue size: 0

Live output at: http://107.22.172.223:8877/52cacf1685a64fa/output.txt

pdfjsbot · 2014-04-29T15:45:54Z

From: Bot.io (Linux)

Received

Command cmd_test from @Snuffleupagus received. Current queue size: 0

Live output at: http://107.21.233.14:8877/fbbf53ccc33c842/output.txt

pdfjsbot · 2014-04-29T16:08:06Z

From: Bot.io (Windows)

Failed

Full output at http://107.22.172.223:8877/52cacf1685a64fa/output.txt

Total script time: 22.17 mins

Font tests: Passed
Unit tests: Passed
Regression tests: FAILED

Image differences available at: http://107.22.172.223:8877/52cacf1685a64fa/reftest-analyzer.html#web=eq.log

pdfjsbot · 2014-04-29T16:12:36Z

From: Bot.io (Linux)

Failed

Full output at http://107.21.233.14:8877/fbbf53ccc33c842/output.txt

Total script time: 26.70 mins

Font tests: Passed
Unit tests: Passed
Regression tests: FAILED

Image differences available at: http://107.21.233.14:8877/fbbf53ccc33c842/reftest-analyzer.html#web=eq.log

timvandermeij · 2014-04-29T16:56:22Z

Something has gone with with the CalGrayCS refactoring I think, given that all CalGray tests fail.

p01 · 2014-04-30T10:13:36Z

Thanks. Good thing all the CalGray test failed. Made it easier to fix them all :)

p01 · 2014-04-30T10:58:21Z

As usual when working on/optimizing color conversions, there might be rounding errors here and there, but nothing noticeable.

Snuffleupagus · 2014-04-30T14:14:49Z

src/core/jpg.js

-        y = data[i + 2] * 0.00392156862745098;
-        k = data[i + 3] * 0.00392156862745098;
+      var length = data.length;
+      var min = -255 *255 * 255;


Nit: missing space after the first multiplication sign.

p01 · 2014-04-30T14:41:51Z

Thanks!
Any chance to get a run of pdfjsbot ?

yurydelendik · 2014-04-30T15:07:31Z

/botio test

pdfjsbot · 2014-04-30T15:07:32Z

From: Bot.io (Windows)

Received

Command cmd_test from @yurydelendik received. Current queue size: 0

Live output at: http://107.22.172.223:8877/5cc9c869a85b530/output.txt

pdfjsbot · 2014-04-30T15:07:32Z

From: Bot.io (Linux)

Received

Command cmd_test from @yurydelendik received. Current queue size: 0

Live output at: http://107.21.233.14:8877/3ce17d6cbd0b940/output.txt

pdfjsbot · 2014-04-30T15:08:44Z

From: Bot.io (Windows)

Failed

Full output at http://107.22.172.223:8877/5cc9c869a85b530/output.txt

Total script time: 1.19 mins

Font tests: FAILED
Unit tests: FAILED
Regression tests: FAILED

Image differences available at: http://107.22.172.223:8877/5cc9c869a85b530/reftest-analyzer.html#web=eq.log

yurydelendik · 2014-04-30T15:29:12Z

/botio-windows test

pdfjsbot · 2014-04-30T15:29:13Z

From: Bot.io (Windows)

Received

Command cmd_test from @yurydelendik received. Current queue size: 0

Live output at: http://107.22.172.223:8877/0e123f29fdd66a4/output.txt

pdfjsbot · 2014-04-30T15:33:21Z

From: Bot.io (Linux)

Success

Full output at http://107.21.233.14:8877/3ce17d6cbd0b940/output.txt

Total script time: 25.82 mins

Font tests: Passed
Unit tests: Passed
Regression tests: Passed

pdfjsbot · 2014-04-30T15:51:17Z

From: Bot.io (Windows)

Success

Full output at http://107.22.172.223:8877/0e123f29fdd66a4/output.txt

Total script time: 22.05 mins

Font tests: Passed
Unit tests: Passed
Regression tests: Passed

p01 · 2014-04-30T21:06:13Z

Is the commit history OK or should I squash some/all ?

yurydelendik · 2014-04-30T21:13:48Z

I think we are fine, but we might be missing the polyfill for Uint8Array

p01 · 2014-04-30T21:38:47Z

Which polyfill ?

About the history, I just noticed 2 fixups squashed in the wrong commit. I'll tidy this up on Friday, or squash it all to one commit tomorrow, as you prefer.

yurydelendik · 2014-05-01T12:38:17Z

src/core/image.js

+        r = (buffer[i] - matteR) * k + matteR;
+        g = (buffer[i + 1] - matteG) * k + matteG;
+        b = (buffer[i + 2] - matteB) * k + matteB;
+        buffer[i] = r <= 0 ? 0 : r >= 255 ? 255 : r | 0;


I'm thinking moving cast to int32 (r | 0) above and giving r the int32 type might speedup the comparison operations (r <= 0 and r >= 255). I run tests at http://jsperf.com/precast to confirm this

I contributed two test runs on IE11 (listed as Other by JSPerf) besides being significantly slower than real browsers, in IE the "after" code was faster.

p01 · 2014-05-02T08:51:51Z

I just ran Yury's test in Opera 22 and IE 11 on my 2-3yo desktop computer. Casting before hand was 23% and 12% faster respectively.

Thanks for the test. The results do make sense.
We can't ignore a potential 12-23% speed improvement. I will see where pre-casting helps.

…lockData Linearize and scale the mcu blocks into the destination buffer in a single pass

timvandermeij · 2014-05-05T13:38:28Z

/botio test

pdfjsbot · 2014-05-05T13:38:30Z

From: Bot.io (Windows)

Received

Command cmd_test from @timvandermeij received. Current queue size: 0

Live output at: http://107.22.172.223:8877/8038daccdb1bacf/output.txt

pdfjsbot · 2014-05-05T13:38:30Z

From: Bot.io (Linux)

Received

Command cmd_test from @timvandermeij received. Current queue size: 0

Live output at: http://107.21.233.14:8877/f6ceb1d025e62b9/output.txt

pdfjsbot · 2014-05-05T14:00:37Z

From: Bot.io (Windows)

Success

Full output at http://107.22.172.223:8877/8038daccdb1bacf/output.txt

Total script time: 22.10 mins

Font tests: Passed
Unit tests: Passed
Regression tests: Passed

pdfjsbot · 2014-05-05T14:04:09Z

From: Bot.io (Linux)

Success

Full output at http://107.21.233.14:8877/f6ceb1d025e62b9/output.txt

Total script time: 25.65 mins

Font tests: Passed
Unit tests: Passed
Regression tests: Passed

Graphic optimizations 1

yurydelendik · 2014-05-05T15:37:10Z

Thank you for the patch

Snuffleupagus reviewed Apr 30, 2014
View reviewed changes

yurydelendik reviewed May 1, 2014
View reviewed changes

p01 added 4 commits May 5, 2014 14:22

Optimized CalGrayCS's convertToRgb

f70a072

Cleaner putBinaryImageData

83cd530

Optimized PDFImage_undoPreblend

27d532e

Optimized JPX's transformComponents

7067409

p01 added 2 commits May 5, 2014 14:36

Optimized JPG.js - reduced memory usage + 2.16x faster getLinearizedB…

287274d

…lockData Linearize and scale the mcu blocks into the destination buffer in a single pass

Optimized JPG.js - 33% faster convertCmykToRgb

14a2d74

yurydelendik added a commit that referenced this pull request May 5, 2014

Merge pull request #4708 from p01/graphic_optimizations_1

7ac1d6c

Graphic optimizations 1

yurydelendik merged commit 7ac1d6c into mozilla:master May 5, 2014

p01 mentioned this pull request May 20, 2014

This PDF takes almost a full minute to load completely #3809

Open

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Graphic optimizations 1 #4708

Graphic optimizations 1 #4708

p01 commented Apr 29, 2014

Snuffleupagus commented Apr 29, 2014

pdfjsbot commented Apr 29, 2014

pdfjsbot commented Apr 29, 2014

pdfjsbot commented Apr 29, 2014

pdfjsbot commented Apr 29, 2014

timvandermeij commented Apr 29, 2014

p01 commented Apr 30, 2014

p01 commented Apr 30, 2014

Snuffleupagus Apr 30, 2014

p01 commented Apr 30, 2014

yurydelendik commented Apr 30, 2014

pdfjsbot commented Apr 30, 2014

pdfjsbot commented Apr 30, 2014

pdfjsbot commented Apr 30, 2014

yurydelendik commented Apr 30, 2014

pdfjsbot commented Apr 30, 2014

pdfjsbot commented Apr 30, 2014

pdfjsbot commented Apr 30, 2014

p01 commented Apr 30, 2014

yurydelendik commented Apr 30, 2014

p01 commented Apr 30, 2014

yurydelendik May 1, 2014

CodingFabian May 1, 2014

p01 commented May 2, 2014

timvandermeij commented May 5, 2014

pdfjsbot commented May 5, 2014

pdfjsbot commented May 5, 2014

pdfjsbot commented May 5, 2014

pdfjsbot commented May 5, 2014

yurydelendik commented May 5, 2014

Graphic optimizations 1 #4708

Graphic optimizations 1 #4708

Conversation

p01 commented Apr 29, 2014

Snuffleupagus commented Apr 29, 2014

pdfjsbot commented Apr 29, 2014

From: Bot.io (Windows)

Received

pdfjsbot commented Apr 29, 2014

From: Bot.io (Linux)

Received

pdfjsbot commented Apr 29, 2014

From: Bot.io (Windows)

Failed

pdfjsbot commented Apr 29, 2014

From: Bot.io (Linux)

Failed

timvandermeij commented Apr 29, 2014

p01 commented Apr 30, 2014

p01 commented Apr 30, 2014

Snuffleupagus Apr 30, 2014

Choose a reason for hiding this comment

p01 commented Apr 30, 2014

yurydelendik commented Apr 30, 2014

pdfjsbot commented Apr 30, 2014

From: Bot.io (Windows)

Received

pdfjsbot commented Apr 30, 2014

From: Bot.io (Linux)

Received

pdfjsbot commented Apr 30, 2014

From: Bot.io (Windows)

Failed

yurydelendik commented Apr 30, 2014

pdfjsbot commented Apr 30, 2014

From: Bot.io (Windows)

Received

pdfjsbot commented Apr 30, 2014

From: Bot.io (Linux)

Success

pdfjsbot commented Apr 30, 2014

From: Bot.io (Windows)

Success

p01 commented Apr 30, 2014

yurydelendik commented Apr 30, 2014

p01 commented Apr 30, 2014

yurydelendik May 1, 2014

Choose a reason for hiding this comment

CodingFabian May 1, 2014

Choose a reason for hiding this comment

p01 commented May 2, 2014

timvandermeij commented May 5, 2014

pdfjsbot commented May 5, 2014

From: Bot.io (Windows)

Received

pdfjsbot commented May 5, 2014

From: Bot.io (Linux)

Received

pdfjsbot commented May 5, 2014

From: Bot.io (Windows)

Success

pdfjsbot commented May 5, 2014

From: Bot.io (Linux)

Success

yurydelendik commented May 5, 2014