The regression was due to the fact that we were calling PlatformColorProc() for
every span (which in turns makes CPUID, a fairly expensive call). Since we draw
a lot of rects, and rects have 1-pixel wide spans for the vertical segments,
that's a lot of CPUID.
Fixed by cacheing the result of PlatformColorProc(), as is done for the other
platform-specific blitters.
Review URL: http://codereview.appspot.com/3669042/
git-svn-id: http://skia.googlecode.com/svn/trunk@636 2bbb7eff-a529-9590-31e7-b0007b416f81
[Patch from weiwei.li@intel.com]
SSE2 optimization has been added by Stephen White before, this improves the skia
performance on SSE2-supporting platform. (please refer to below issues)
Issue 171055: More SSE2ification
Issue 157141: More SSE2ification
Issue 150060: minor tweaks to SSE2 code for -fPIC
Issue 144072: SSE2 optimizations for 32bit blending blitters
This CL implements SSE2 optimizations for the 32bit Color operation. Like above
issues, it uses CPUID to detect for SSE2 and changes the platform procs at
runtime as well. The 32bit Color operation is heavily used on Chrome HTML5
canvas operations. Take Microsoft IE test drives Pulsating Bubbles as example
(http://ie.microsoft.com/testdrive/Performance/PulsatingBubbles/Default.xhtml),
if running this cases on Chrome, the overhead of 32bit Color operation is about
40~50%. So this CL will make skia performance more better, and also make Chrome
HTML5 canvas performance more better.
Additional, this CL has passed the skia bench & tests validation, the result is
pretty good. We also apply this CL to the latest chromium, and re-run Microsoft
IE test drives Pulsating Bubbles, the performance is improved by almost 9~10%.
git-svn-id: http://skia.googlecode.com/svn/trunk@633 2bbb7eff-a529-9590-31e7-b0007b416f81
This patch adds four methods to SkBitmap. There are two functions to return
"safe size", defined as the number of pixels from the value returned by
getPixels() to the end of the allocated buffer.
There is one version of fillPixels() to copy the bitmap instance into an
external buffer (with specified size, and using specified stride), and another
fillPixels() to copy from an external buffer to the instance bitmap. In the
latter case the specified height, width and pixel format must match that used by
the bitmap instance, although the specified stride may be any value at least as
large as the minimum stride for the specified geometry. It is assumed that the
external buffer is of size at least (height - 1)*stride + width *
bytesPerPixel.
Both fillPixels() functions return false if the copy is not possible with the
specified parameters.
Review URL: http://codereview.appspot.com/2837041/
git-svn-id: http://skia.googlecode.com/svn/trunk@625 2bbb7eff-a529-9590-31e7-b0007b416f81
Supports fakeBold, underline, strikethrough, mode (fill, stroke, both), size, skew, alignment (left, center, right).
Missing is drawFontOnPath and font lookup and embedding.
Changed SkPDFString to support how it is used from drawText methods.
Moved compile assert into SkTypes.
Moved constants and utility function used to support fakeBold, underline, and strikethrough into higher level locations.
Review URL: http://codereview.appspot.com/2946041
git-svn-id: http://skia.googlecode.com/svn/trunk@624 2bbb7eff-a529-9590-31e7-b0007b416f81
The '#include "SkGLDevice.h"' from include/core/SkDevice.h requires internal
Skia code to be added to the include search paths when using the deprecated API.
This change adds back SkGLCanvas.cpp to avoid exposing SkGLDevice.h to the
public API.
The change also includes an explicit virtual destructor on SkDeviceFactory to
silence a -Wnon-virtual-dtor warning and allow for -Werror.
Original Issue: http://codereview.appspot.com/3009041/
Review URL: http://codereview.appspot.com/2994042
git-svn-id: http://skia.googlecode.com/svn/trunk@623 2bbb7eff-a529-9590-31e7-b0007b416f81
Reorganize how the PDF graphic state stack is managed (fixing several bugs incidentally).
Style: fix variables with underscores.
Bug: fix image matrix application order, which enabled a small refactor.
Review URL: http://codereview.appspot.com/2771042
git-svn-id: http://skia.googlecode.com/svn/trunk@622 2bbb7eff-a529-9590-31e7-b0007b416f81
Fix the spelling of resource in several places
Make getResouce(resourceList) part of SkPDFObject
make SkDynamicMemoryWStream::getOffset and SkPDFPage::getMediaBox const
Add a temporary NOT_IMPLEMENTED macro instead of using SkASSERT
Review URL: http://codereview.appspot.com/2721041
git-svn-id: http://skia.googlecode.com/svn/trunk@619 2bbb7eff-a529-9590-31e7-b0007b416f81
These are not the capabilities of the factory, but of the device. Additionally, it is more often needed when you have a device then when you have a factory, which caused creating of a new factory.
Review URL: http://codereview.appspot.com/2720041
git-svn-id: http://skia.googlecode.com/svn/trunk@618 2bbb7eff-a529-9590-31e7-b0007b416f81
The previous change made it difficult to inherit from SkCanvas without leaking memory. By making SkDeviceFactory not reference counted, the right thing happens more naturally, just NewCanvas : public SkCanvas(new NewDeviceFactory()) {...}
Review URL: http://codereview.appspot.com/2530042
git-svn-id: http://skia.googlecode.com/svn/trunk@605 2bbb7eff-a529-9590-31e7-b0007b416f81
Methods or classes that should go away are marked deprecated. The only thing I know of that breaks backward compatibility is SkCanvas((SkDevice*)NULL), but that is fairly unlikely to occur in the wild because that constructor had a default value of NULL.
Review URL: http://codereview.appspot.com/2103045
git-svn-id: http://skia.googlecode.com/svn/trunk@604 2bbb7eff-a529-9590-31e7-b0007b416f81
- when the previous point in the path and (x1, y1) are coincident
- when (x1, y1) and (x2, y2) are coincident
These cause the setNormalize() calls to leave the "before" and "after" vectors
uninitialized, respectively, which are then used to compute cosh and sinh.
I chose to implement HTML5 canvas behaviour for these cases (as well as for the
case where all three points are collinear): add a line to the first point, and
bail. I think this is ok, since skia is mostly used for HTML5 browsers, and
this case is degenerate anyway.
Review URL: http://codereview.appspot.com/2484041
git-svn-id: http://skia.googlecode.com/svn/trunk@603 2bbb7eff-a529-9590-31e7-b0007b416f81
have a bitmap with pixels, but no (required) colortable. readyToDraw() knows
to check for this.
git-svn-id: http://skia.googlecode.com/svn/trunk@599 2bbb7eff-a529-9590-31e7-b0007b416f81
The S32A_Blend_BlitRow32 function was written and unrolled in ARM
assembly to improve the rendering performance on ARM cores lacking
NEON.
Added in SkBlitRow_opts_arm.cpp.
Performance improvement about ~18% on micro benchmarking.
Patch-by: Vassillis
git-svn-id: http://skia.googlecode.com/svn/trunk@598 2bbb7eff-a529-9590-31e7-b0007b416f81
inside SkBitmap.cpp.
core_files.mk updated, but if your build does not reference that, you will
need to add SkMallocPixelRef.cpp to your list
git-svn-id: http://skia.googlecode.com/svn/trunk@597 2bbb7eff-a529-9590-31e7-b0007b416f81
Flagged as a compiler warning. It appears that it was intended that the
conjuction not bind so tightly.
git-svn-id: http://skia.googlecode.com/svn/trunk@591 2bbb7eff-a529-9590-31e7-b0007b416f81
