'GPU' is in 'Test-Ubuntu14-GCE-NoGPU-x86_64-Release-Valgrind_CPU' too.
This means we're building it in no-GPU mode, and running it in no-CPU mode.
At least it finishes quite quickly this way (~10 seconds).
BUG=skia:
Review URL: https://codereview.chromium.org/992203004
Pictures with many drawn paths benefit from the increase of the
count. As an example, chalkboard testcase has ~2500 paths, tiger
has ~300.
Bigger objects, like textures, tend to be limited by the cache
byte limit, not item count.
Review URL: https://codereview.chromium.org/991783003
This changes SkScalarClampMax and SkScalarPin to use SkTMin and SkTMax.
This change allows compilers to more easily transform these operations
into fast max/min operarations as opposed to conditional branches.
Review URL: https://codereview.chromium.org/993593002
Reason for revert:
There's some weirdness with Tegra2 and Tegra3 GMs (Xoom and Nexus7) that may be related to this.
Original issue's description:
> Adjust distance field glyph and font atlas sizes to maximize storage.
>
> Because of high DPI devices, we need more room in the glyph
> atlas for the larger glyphs. These settings will allow 4 of
> the distance field glyphs to fit in one Plot (increasing
> the storage from 32 large glyphs to 128), and still permit
> us to handle glyphs up to a 312 point size.
>
> BUG=chromium:458791
>
> Committed: https://skia.googlesource.com/skia/+/d2737ad7dd8f4ea94a74034027014fd3d78923cbTBR=joshualitt@google.com,bsalomon@google.com
NOPRESUBMIT=true
NOTREECHECKS=true
NOTRY=true
BUG=chromium:458791
Review URL: https://codereview.chromium.org/986333005
Because of high DPI devices, we need more room in the glyph
atlas for the larger glyphs. These settings will allow 4 of
the distance field glyphs to fit in one Plot (increasing
the storage from 32 large glyphs to 128), and still permit
us to handle glyphs up to a 312 point size.
BUG=chromium:458791
Review URL: https://codereview.chromium.org/986343002
Reason for revert:
Lots of PDF failures -- need to investigate
Original issue's description:
> use colorfilter instead of imagefilter when possible
>
> big speed win w/ nanobench --match colorfilter_
>
> BUG=skia:
> TBR=
>
> Will require rebaselining, so landing temporarily to trigger layouttests.
>
> Committed: https://skia.googlesource.com/skia/+/f641bd076d29666a5d2a10d5384e4924b0a0e524TBR=sugoi@google.com,senorblanco@google.com,bsalomon@google.com,reed@google.com,senorblanco@chromium.org
NOPRESUBMIT=true
NOTREECHECKS=true
NOTRY=true
BUG=skia:
Review URL: https://codereview.chromium.org/991693002
Clarify asColorFilter ...
1. Rename to isColorFilterNode for DAG reduction
2. Add asAColorFilter for removing the imagefilter entirely (future use-case)
Need layouttest rebaseline suppression before this can land in chrome...
https://codereview.chromium.org/984023004/
BUG=skia:
Review URL: https://codereview.chromium.org/982933002
Reason for revert:
Reverting in case this is the cause of the non-Windows failures.
Original issue's description:
> For consistency, use our homebrew zlib everywhere possible.
>
> This switches when we build our own zlib from "just Windows" to "everyone, but
> not Android framework of course".
>
> I tested this by building DM for my Mac and for an Android bot config.
> It took minor tweaks to the GYP to get ARM builds working.
>
> BUG=skia:
>
> Committed: https://skia.googlesource.com/skia/+/5a8f2257b0b0f954fb74f65e7ea3ada772ed9240TBR=scroggo@google.com,mtklein@chromium.org
NOPRESUBMIT=true
NOTREECHECKS=true
NOTRY=true
BUG=skia:
Review URL: https://codereview.chromium.org/989873002
When computed, the RTree for an SkPicture will have a root
bounds that reflects the best bounding information available,
rather than the best estimate at the time the picture recorder
is created. Given that creators frequently don't know ahead of
time what will be drawn, the RTree bound is often tighter.
Perf testing on Chrome indicates a small raster performance
advantage. For upcoming painting changes in Chrome the
performance advantage is much larger.
BUG=
Committed: https://skia.googlesource.com/skia/+/2dd3b6647dc726f36fd8774b3d0d2e83b493aeac
Review URL: https://codereview.chromium.org/971803002
This switches when we build our own zlib from "just Windows" to "everyone, but
not Android framework of course".
I tested this by building DM for my Mac and for an Android bot config.
It took minor tweaks to the GYP to get ARM builds working.
BUG=skia:
Review URL: https://codereview.chromium.org/971673005
The tests assert on Sk_ColorGREEN, which is in BRGA non-premul. Read the
pixels in same color format.
Currently the tests pass on all platforms because GREEN is fully opaque
and the component stays in the same place both on BGRA and
RGBA. However, hypothetically somebody could copy-paste the assertion
for further tests but use, say, RED. This creates latent error that is
only visible on some platforms like mac.
Review URL: https://codereview.chromium.org/989463002
Clamping 4 at a time is now about 15% faster than 1 at a time with SSSE3.
Clamping 4 at a time is now about 20% faster with SSE2,
and this applies to non-clamping too (we still just clamp there).
In all cases, 4 at a time is never worse than 1 at a time,
and not clamping is never slower than clamping.
Here's all the bench results, with the numbers for portable code as a fun point
of reference:
SSSE3:
maxrss loops min median mean max stddev samples config bench
10M 2291 4.66ns 4.66ns 4.66ns 4.68ns 0% ▆█▁▁▁▇▁▇▁▃ nonrendering SkPMFloat_get_1x
10M 2040 5.29ns 5.3ns 5.3ns 5.32ns 0% ▃▆▃▃▁▁▆▃▃█ nonrendering SkPMFloat_clamp_1x
10M 7175 4.62ns 4.62ns 4.62ns 4.63ns 0% ▁▄▃████▃▄▇ nonrendering SkPMFloat_get_4x
10M 5801 4.89ns 4.89ns 4.89ns 4.91ns 0% █▂▄▃▁▃▄█▁▁ nonrendering SkPMFloat_clamp_4x
SSE2:
maxrss loops min median mean max stddev samples config bench
10M 1601 6.02ns 6.05ns 6.04ns 6.08ns 0% █▅▄▅▄▂▁▂▂▂ nonrendering SkPMFloat_get_1x
10M 2918 6.05ns 6.06ns 6.05ns 6.06ns 0% ▂▇▁▇▇▁▇█▇▂ nonrendering SkPMFloat_clamp_1x
10M 3569 5.43ns 5.45ns 5.44ns 5.45ns 0% ▄█▂██▇▁▁▇▇ nonrendering SkPMFloat_get_4x
10M 4168 5.43ns 5.43ns 5.43ns 5.44ns 0% █▄▇▁▇▄▁▁▁▁ nonrendering SkPMFloat_clamp_4x
Portable:
maxrss loops min median mean max stddev samples config bench
10M 500 27.8ns 28.1ns 28ns 28.2ns 0% ▃█▆▃▇▃▆▁▇▂ nonrendering SkPMFloat_get_1x
10M 770 40.1ns 40.2ns 40.2ns 40.3ns 0% ▅▁▃▂▆▄█▂▅▂ nonrendering SkPMFloat_clamp_1x
10M 1269 28.4ns 28.8ns 29.1ns 32.7ns 4% ▂▂▂█▂▁▁▂▁▁ nonrendering SkPMFloat_get_4x
10M 1439 40.2ns 40.4ns 40.4ns 40.5ns 0% ▆▆▆█▁▆▅█▅▆ nonrendering SkPMFloat_clamp_4x
SkPMFloat_neon.h is still one big TODO as far as 4-at-a-time APIs go.
BUG=skia:
Review URL: https://codereview.chromium.org/982123002
Reason for revert:
Need to suppress these for rebaselining, so reverting for now.
Regressions: Unexpected image-only failures (5)
css3/filters/effect-brightness-clamping-hw.html [ ImageOnlyFailure ]
css3/filters/effect-combined-hw.html [ ImageOnlyFailure ]
virtual/slimmingpaint/css3/filters/effect-brightness-clamping-hw.html [ ImageOnlyFailure ]
virtual/slimmingpaint/css3/filters/effect-combined-hw.html [ ImageOnlyFailure ]
Original issue's description:
> Use ComposeColorFilter in factory to collapse consecutive filters (when possible).
> Change asColorFilter to reflect its reliance on the new factory behavior.
>
> patch from issue 967143002 at patchset 80001 (http://crrev.com/967143002#ps80001)
>
> BUG=skia:
>
> Committed: https://skia.googlesource.com/skia/+/dac843bf046c2cd79fd955cb177aee241d7a4b0cTBR=senorblanco@chromium.org,robertphillips@google.com,bsalomon@chromium.org
NOPRESUBMIT=true
NOTREECHECKS=true
NOTRY=true
BUG=skia:
Review URL: https://codereview.chromium.org/978923005
This blacklist entry bans any test with 'pdf' config, any source type, whose
name has '.webp' in it. In practice, that's 'image' or 'subset' source type
decoding some WEBP file.
BUG=skia:3505
Review URL: https://codereview.chromium.org/982163002
Reason for revert:
Might be breaking deps roll
Original issue's description:
> Update SkPicture cull rects with RTree information
>
> When computed, the RTree for an SkPicture will have a root
> bounds that reflects the best bounding information available,
> rather than the best estimate at the time the picture recorder
> is created. Given that creators frequently don't know ahead of
> time what will be drawn, the RTree bound is often tighter.
>
> Perf testing on Chrome indicates a small raster performance
> advantage. For upcoming painting changes in Chrome the
> performance advantage is much larger.
>
> BUG=
>
> Committed: https://skia.googlesource.com/skia/+/2dd3b6647dc726f36fd8774b3d0d2e83b493aeacTBR=mtklein@google.com,schenney@chromium.org
NOPRESUBMIT=true
NOTREECHECKS=true
NOTRY=true
BUG=
Review URL: https://codereview.chromium.org/977413003
The core of the problem is that the system is asked to lookup the metrics for a character with id == 0. This causes a hit in the fCharToGlyphHash matching the sentinel glyph. This happens because fCharToGlpyhHash is initialized with all zeros, therefore, the fID is zero matching the char with id == 0. The fAdvanceX field of the sentinel glyph is in fact not initialized.
The bigger question is now did a zero character get passed to getUnicharMetrics?
The breaking code is basically as follows:
wchar_t glyph = L'S';
paint.measureText(&glyph, 2);
This get mischaracterized as a utf8 string instead of a utf16(?) string. Because of the little endian ordering, this is the character string 'L' '\0'. Since the size of the original string is two bytes (but a single character) the '\0' is treated as its own character and past to getUnicharMetrics.
TEST:
On windows failed using DrMemory. With this change does not fail.
BUG=463204
Review URL: https://codereview.chromium.org/977063002
Also allow incomplete to be considered successful.
Do not attempt to draw transparent images on 565.
BUG=skia:3475
Review URL: https://codereview.chromium.org/978413002
Please see if this looks usable. It may even give a perf boost if you use it, even without custom implementations for each instruction set.
I've been trying this morning to beat this naive loop implementation, but so far no luck with either _SSE2.h or _SSSE3.h. It's possible this is an artifact of the microbenchmark, because we're not doing anything between the conversions. I'd like to see how this fits into real code, what assembly's generated, what the hot spots are, etc.
I've updated the tests to test these new APIs, and splintered off a pair of new benchmarks that use the new APIs. This required some minor rejiggering in the benches.
BUG=skia:
Review URL: https://codereview.chromium.org/978213003
The main thrust of this CL is to remove the currCmdMarker variable from GrTargetCommands::flush. In a MultiDrawBuffer world the Cmds need not be execute in the order of their issuance.
Review URL: https://codereview.chromium.org/978363002
Tasks that produce a non-fatal error will bail out before writing their output to
disk and hash to dm.json, but not count as failures.
This also makes true failures bail out before writing their results. If the DM
program failed, we probably don't want to triage that image result.
We use this new feature first to skip image subset decoding when we detect it's
not supported. Here's a snippet of an example run, where in this case only
.webp are subset decodable:
...
( 15MB 12) 172µs 8888 subset color_wheel.jpg (skipped: Subset decoding not supported.)
( 15MB 11) 9.05ms 8888 subset randPixels.webp
( 16MB 10) 863µs 8888 subset baby_tux.png (skipped: Subset decoding not supported.)
...
Only outputs corresponding to the .webp show up, both on disk and in the .json.
BUG=skia:
Review URL: https://codereview.chromium.org/980333002
