Reason for revert:
breaks build when not using SSE3, since the two method definitions differ in parameter types (typo)
Original issue's description:
> Add SSSE3 acceleration for S32_D16_filter_DXDY
>
> With this CL, related nanobench can be improved for 565 config.
> bitmap_BGRA_8888_scale_rotate_bilerp 115us -> 70.5us 0.61x
> bitmap_BGRA_8888_update_volatile_scale_rotate_bilerp 115us -> 70.5us 0.61x
> bitmap_BGRA_8888_update_scale_rotate_bilerp 112us -> 68us 0.6x
>
>
> BUG=skia:
>
> Committed: https://skia.googlesource.com/skia/+/45a05780867a06b9f8a8d5240cf6c5d5a2c15a35TBR=mtklein@google.com,qkmiao@gmail.com,qiankun.miao@intel.com
NOTREECHECKS=true
NOTRY=true
BUG=skia:
Review URL: https://codereview.chromium.org/761103003
Seems okay after this small patch to skip lockPixels() / unlockPixels().
BUG=skia:3149
CQ_EXTRA_TRYBOTS=client.skia:Test-Ubuntu13.10-GCE-NoGPU-x86_64-Release-TSAN-Trybot
Review URL: https://codereview.chromium.org/773203003
We do alias src and dst when building the BBH. This meant we calculated
the wrong bounds in LayoutTests/ietestcenter/css3/text/textshadow-010.htm.
Not sure how best to test this inside Skia. It may become moot after I
work through the API changes we were talking about today, but if not I'll
make sure it's tested somehow before I'm done with SkDrawLoopers.
BUG=skia:3186
Review URL: https://codereview.chromium.org/760023004
I got my hands on a list of new warnings coming out of MSVC when run with /analyze.
I scanned over them and didn't see any bugs, but I did notice this is just a weird
way to write this expression. (MSVC /analyze was wary of '=' in a boolean expression.)
BUG=skia:
Review URL: https://codereview.chromium.org/776033002
The debugger accepts either a single skp or a directory as a parameter.
The webpages_playback script used to pass multiple files using a '*'
glob.
Review URL: https://codereview.chromium.org/775073002
In python, str.lstrip([chars]) means "lstrip any of the chars from str",
not "remove prefix". Thus "gs://skia".lstrip("gs://") yields "kia".
Replace with custom function stripping the prefix from lstrip()ed gs
base name.
Review URL: https://codereview.chromium.org/780493002
The render_pdfs nowadays seems to input the skps from
the dir passed as '-r' parameter.
The app will not output anything, but it will not crash the
webpages_playback script either. Affects only runs that have
--skia_tools=<dir> argument.
Review URL: https://codereview.chromium.org/776873004
With this CL, related nanobench can be improved for 565 config.
bitmap_BGRA_8888_scale_rotate_bilerp 115us -> 70.5us 0.61x
bitmap_BGRA_8888_update_volatile_scale_rotate_bilerp 115us -> 70.5us 0.61x
bitmap_BGRA_8888_update_scale_rotate_bilerp 112us -> 68us 0.6x
BUG=skia:
Review URL: https://codereview.chromium.org/773753002
When checking whether we need apply AA to a rect for non-msaa target,
simply checking that the mapped rect is integer is not enough.
We need to check whether the transformation matrix has transformations
other than simple translation.
Original-Author: Henry Song <henrysong@samsung.com>
Review URL: https://codereview.chromium.org/772953002
If the dest isn't pixel aligned, or if a non 90 degree rotation is
present, we need to use drawRect() for drawing anti-aliased bitmaps on
non-msaa targets or the edges won't be anti-aliased as intended.
Note: If the bitmap size is bigger than max texture size we fall back to drawBitmapCommon.
Original-Author: Henry Song <henrysong@samsung.com>
Review URL: https://codereview.chromium.org/649313003
Mostly this means using SkLazyPtr for the 16-bit cache.
We can remove the copy constructor now and just ref it instead.
BUG=skia:
Review URL: https://codereview.chromium.org/769323002
Motivation: colorwheel on pdf-native (MacOS coregraphics) reverses Blue and Red. I want to see if this has anything to do with image decoding.
BUG=skia:3184
Review URL: https://codereview.chromium.org/774713003
