blink skips all pending commands during picture recording if it is drawing an opaque full-frame
geometry or image. This may improve performance for some edge cases. To recognize an opaque
full-frame drawing should be cheap enough. Otherwise, the overhead will offset the improvement.
Unfortunately, data from perf for content_shell on Nexus7 shows that SkDeferredCanvas::isFullFrame
is far from cheap. Table below shows that how much isFullFrame() costs in the whole render process.
benchmark percentage
my local benchmark(draw 1000 sprites) 4.1%
speedReading 2.8%
FishIETank(1000 fishes) 1.5%
GUIMark3 Bitmap 2.0%
By contrast, real recording (SkGPipeCanvas::drawBitmapRectToRect) and real rasterization
(GrDrawTarget::drawRect) cost ~4% and ~6% in the whole render process respectively. Apparently,
SkDeferredCanvas::isFullFrame() is nontrivial.
getDeviceSize() is the main contributor to this hotspot. The change simply save the canvasSize and
reuse it among drawings if it is not a fresh frame. This change cut off ~65% (or improved ~2 times)
of isFullFrame().
telemetry smoothness canvas_tough_test didn't show obvious improvement or regression.
BUG=411166
Committed: https://skia.googlesource.com/skia/+/8e45c3777d886ba3fe239bb549d06b0693692152R=junov@chromium.org, tomhudson@google.com, reed@google.com
Author: yunchao.he@intel.com
Review URL: https://codereview.chromium.org/545813002
Reason for revert:
This is leaking memory:
http://108.170.220.120:10117/builders/Test-Ubuntu13.10-GCE-NoGPU-x86_64-Debug-ASAN/builds/2516/steps/RunDM/logs/stdio
Original issue's description:
> Picture Recording: fix the performance bottleneck in SkDeferredCanvas::isFullFrame
>
> blink skips all pending commands during picture recording if it is drawing an opaque full-frame
> geometry or image. This may improve performance for some edge cases. To recognize an opaque
> full-frame drawing should be cheap enough. Otherwise, the overhead will offset the improvement.
> Unfortunately, data from perf for content_shell on Nexus7 shows that SkDeferredCanvas::isFullFrame
> is far from cheap. Table below shows that how much isFullFrame() costs in the whole render process.
>
> benchmark percentage
> my local benchmark(draw 1000 sprites) 4.1%
> speedReading 2.8%
> FishIETank(1000 fishes) 1.5%
> GUIMark3 Bitmap 2.0%
>
> By contrast, real recording (SkGPipeCanvas::drawBitmapRectToRect) and real rasterization
> (GrDrawTarget::drawRect) cost ~4% and ~6% in the whole render process respectively. Apparently,
> SkDeferredCanvas::isFullFrame() is nontrivial.
>
> getDeviceSize() is the main contributor to this hotspot. The change simply save the canvasSize and
> reuse it among drawings if it is not a fresh frame. This change cut off ~65% (or improved ~2 times)
> of isFullFrame().
>
> telemetry smoothness canvas_tough_test didn't show obvious improvement or regression.
>
> BUG=411166
>
> Committed: https://skia.googlesource.com/skia/+/8e45c3777d886ba3fe239bb549d06b0693692152R=junov@chromium.org, tomhudson@google.com, reed@google.com, yunchao.he@intel.comTBR=junov@chromium.org, reed@google.com, tomhudson@google.com, yunchao.he@intel.com
NOTREECHECKS=true
NOTRY=true
BUG=411166
Author: mtklein@google.com
Review URL: https://codereview.chromium.org/571053002
blink skips all pending commands during picture recording if it is drawing an opaque full-frame
geometry or image. This may improve performance for some edge cases. To recognize an opaque
full-frame drawing should be cheap enough. Otherwise, the overhead will offset the improvement.
Unfortunately, data from perf for content_shell on Nexus7 shows that SkDeferredCanvas::isFullFrame
is far from cheap. Table below shows that how much isFullFrame() costs in the whole render process.
benchmark percentage
my local benchmark(draw 1000 sprites) 4.1%
speedReading 2.8%
FishIETank(1000 fishes) 1.5%
GUIMark3 Bitmap 2.0%
By contrast, real recording (SkGPipeCanvas::drawBitmapRectToRect) and real rasterization
(GrDrawTarget::drawRect) cost ~4% and ~6% in the whole render process respectively. Apparently,
SkDeferredCanvas::isFullFrame() is nontrivial.
getDeviceSize() is the main contributor to this hotspot. The change simply save the canvasSize and
reuse it among drawings if it is not a fresh frame. This change cut off ~65% (or improved ~2 times)
of isFullFrame().
telemetry smoothness canvas_tough_test didn't show obvious improvement or regression.
BUG=411166
R=junov@chromium.org, tomhudson@google.com, reed@google.com
Author: yunchao.he@intel.com
Review URL: https://codereview.chromium.org/545813002
drawPatch now receives as parameter const SkPoint cubics[12]
Adjusted derived classes and serialization.
Ajusted GM's and benches that take into account combinations of optional
parameters, the scale of the patch and 4 different types of patches.
Planning on adding the extra functionality of SkPatch in another CL.
BUG=skia:
R=egdaniel@google.com, reed@google.com
Author: dandov@google.com
Review URL: https://codereview.chromium.org/463493002
Added function SkCanvas::drawPatch to the API. This function
receives the patch to draw and the paint.
Added function SkBaseDevice::drawPatch to the API. This function also receives the patch to draw and the paint.
Currently SkGpuDevice and SkBitmapDevice generate the mesh taking into
account the scale factor and call the corresponding device's drawVertices.
BUG=skia:
R=jvanverth@google.com, egdaniel@google.com, bsalomon@google.com
Author: dandov@google.com
Review URL: https://codereview.chromium.org/424663006
The base class, SkCanvasState, now holds the version, width, and
height. These fields will always be a necessary part of the class.
(Also add in some padding.)
The other fields, which may change, have been moved into the
subclass, SkCanvasState_v1. If/when the version changes, it will
correspond to a new subclass.
In SkCanvasStateUtils::CreateFromCanvasState, check the version on
the base class, then do a static_cast to the version corresponding
to SkCanvasState::version.
Remove CANVAS_STATE_VERSION, which is redundant with the version
specified by the subclass.
Use unambiguous type for rowBytes.
Build Android with SK_SUPPORT_LEGACY_CLIPTOLAYERFLAG. This allows us
to run the full suite of CanvasState tests. It is also representative
of what will be used on Android by WebView.
Fix CanvasStateTest where it was broken inside ifdef'ed out code.
Use SkCanvas::getBaseLayerSize() instead of the deprecated
SkCanvas::getDeviceSize().
Update the comments in the header to be more clear. In particular,
an SkCanvasState can only be used to pass an SkCanvas' state to a
future version of Skia (or the same); not an older version.
NOTREECHECKS=true
BUG=b/15693384
R=reed@google.com, mtklein@google.com, djsollen@google.com
Author: scroggo@google.com
Review URL: https://codereview.chromium.org/372003002
This splits the playback functionality out of SkPictureData. The old SkPictureData::draw method is pulled out along
with its supporting functions as verbatim as possible. Some follow on CLs will be required to:
re-enable profiling in the debugger (and remove the vestiges of SkTimedPicture)
re-enable display of command offsets in the picture (this should probably wait until we've switched to SkRecord though)
Clean up CachedOperationList (maybe fuse with SkPicture::OperationList)
Split SkPicturePlayback into a base class and two derived classes
Implement parallel version of GatherGPUInfo for SkRecord
Landing this is blocked on removing Android's use of the abortPlayback entry point.
R=mtklein@google.com, reed@google.com
Author: robertphillips@google.com
Review URL: https://codereview.chromium.org/377623002
Always build the tools with JSON, but either build our own
or use the system's.
Rename skia_build_json_writer to skia_use_system_jsoncpp,
since we now always build with JSON.
Remove SK_BUILD_JSON_WRITER, which was only there so
we could build without JSON it in the framework.
BUG=skia:2448
R=djsollen@google.com, reed@google.com
Author: scroggo@google.com
Review URL: https://codereview.chromium.org/303913002
This is a follow up to https://codereview.chromium.org/338913002/. More
tweaks to allow migrating clients to the new API.
1) switch the shim call direction willSave(SaveFlags) -> willSave()
(internal users are still using the former, so external overriders of
the latter will not be notified otherwise - doh)
2) ensure willSave() stays visible in SkProxyCanvas (Chromium's
TimingCanvas attempts to call it explicitly).
BUG=skia:2297
R=reed@google.com
Author: fmalita@chromium.org
Review URL: https://codereview.chromium.org/334393008
For now, remove json functionality and do not depend on json. This
allows us to build and run until solving skbug.com/2448.
bench/DeferredSurfaceCopyBench.cpp:
Include GrRenderTarget last, so SK_SUPPORT_GPU will be set properly.
bench/ResultsWriter.h:
bench/benchmain.cpp:
Remove JSONResultsWriter when SK_BUILD_JSON_WRITER is not defined,
which is the case for the Android framework build.
gyp/bench.gyp:
Depend on skia and cutils (for android_atomic_inc etc).
gyp/common_conditions.gypi:
Define SK_BUILD_JSON_WRITER when skia_build_json_writer is set.
gyp/common_variables.gypi:
Add a flag for skia_build_json_writer, and set it only when
skia_android_framework is not set.
gyp/jsoncpp.gyp:
Do not build jsoncpp when skia_build_json_writer is not defined.
include/utils/SkJSONCPP.h:
Do not include json headers when SK_BUILD_JSON_WRITER is not defined.
platform_tools/android/bin/gyp_to_android.py:
Generate bench/Android.mk.
platform_tools/android/gyp_gen/gypd_parser.py:
Skip dest_dir when checking for include_dirs.
platform_tools/android/gyp_gen/makefile_writer.py:
Build bench/Android.mk when building external/skia.
platform_tools/android/gyp_gen/tool_makefile_writer.py:
Add a parameter for putting the binary into /data/local/tmp.
BUG=skia:2447
BUG=skia:2448
R=halcanary@google.com, reed@google.com
Author: scroggo@google.com
Review URL: https://codereview.chromium.org/282053002
git-svn-id: http://skia.googlecode.com/svn/trunk@14760 2bbb7eff-a529-9590-31e7-b0007b416f81
Like yesterday's change to run CPU-parent child tasks serially in thread, this
reduces peak memory usage by improving the temporaly locality of the bitmaps we
create.
E.g. Let's say we start with tasks A B C and D
Queue: [ A B C D ]
Running A creates A' and A", which depend on a bitmap created by A.
Queue: [ B C D A' A" * ]
That bitmap now needs sit around in RAM while B C and D run pointlessly and can
only be destroyed at *. If instead we do this and push dependent child tasks
to the front of the queue, the queue and bitmap lifetime looks like this:
Queue: [ A' A" * B C D ]
This is much, much worse in practice because the queue is often several thousand
tasks long. 100s of megs of bitmaps can pile up for 10s of seconds pointlessly.
To make this work we add addNext() to SkThreadPool and its cousin DMTaskRunner.
I also took the opportunity to swap head and tail in the threadpool
implementation so it matches the comments and intuition better: we always pop
the head, add() puts it at the tail, addNext() at the head.
Before
Debug: 49s, 1403352k peak
Release: 16s, 2064008k peak
After
Debug: 49s, 1234788k peak
Release: 15s, 1903424k peak
BUG=skia:2478
R=bsalomon@google.com, borenet@google.com, mtklein@google.com
Author: mtklein@chromium.org
Review URL: https://codereview.chromium.org/263803003
git-svn-id: http://skia.googlecode.com/svn/trunk@14506 2bbb7eff-a529-9590-31e7-b0007b416f81
