| Commit message (Collapse) | Author | Age |
|
|
|
|
|
|
|
|
|
|
|
|
| |
This should RVO to the same as doing it on the stack with setConcat.
BUG=skia:
R=reed@google.com, mtklein@google.com
Author: mtklein@chromium.org
Review URL: https://codereview.chromium.org/293693002
git-svn-id: http://skia.googlecode.com/svn/trunk@14782 2bbb7eff-a529-9590-31e7-b0007b416f81
|
|
|
|
|
|
|
|
|
|
|
| |
BUG=skia:2378
R=reed@google.com, mtklein@google.com, robertphillips@google.com
Author: mtklein@chromium.org
Review URL: https://codereview.chromium.org/290883004
git-svn-id: http://skia.googlecode.com/svn/trunk@14778 2bbb7eff-a529-9590-31e7-b0007b416f81
|
|
|
|
|
|
|
|
|
|
|
| |
BUG=skia:
R=robertphillips@google.com, mtklein@google.com
Author: mtklein@chromium.org
Review URL: https://codereview.chromium.org/287953002
git-svn-id: http://skia.googlecode.com/svn/trunk@14743 2bbb7eff-a529-9590-31e7-b0007b416f81
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
This does render_pictures, plus checks SkRecord optimizations.
Disable an SkRecord optimization that draws several bot SKPs wrong. (To be investigated.)
BUG=skia:2378
R=reed@google.com, mtklein@google.com
Author: mtklein@chromium.org
Review URL: https://codereview.chromium.org/270543004
git-svn-id: http://skia.googlecode.com/svn/trunk@14739 2bbb7eff-a529-9590-31e7-b0007b416f81
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Prints microsecond timing for each command in the left-hand column:
optimized flat/http___mobile_news_sandbox_google_com_news_pt0_scroll_layer_7.skp
4.0 1 Save
2075.0 2 DrawRect
104.0 3 BoundedDrawPosTextH
135.4 4 DrawRect
9.4 5 DrawRect
5.6 6 DrawRect
8.2 7 DrawRect
6.8 8 DrawRect
...
(I'm sure Rietveld will just mangle the crap out of that. It's helpfully right-aligned.)
To do this, I made Draw from SkRecordDraw Skia-public as SkRecords::Draw,
and time it command-by-command.
BUG=skia:2378
R=fmalita@chromium.org, mtklein@google.com
Author: mtklein@chromium.org
Review URL: https://codereview.chromium.org/272723007
git-svn-id: http://skia.googlecode.com/svn/trunk@14672 2bbb7eff-a529-9590-31e7-b0007b416f81
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Just noticed that the cull noop pattern (PushCull, Star<NoOp>, PopCull)
would trigger the Star that stores matches. We certainly don't need
those matches here, so instead of magically determining which Star you
need, we'll make you tell us which one you want.
No one but List's unit test needs List. I'll leave it for now, but we
might find it's not useful.
BUG=skia:2378
R=fmalita@chromium.org, mtklein@google.com, bungeman@google.com
Author: mtklein@chromium.org
Review URL: https://codereview.chromium.org/275623002
git-svn-id: http://skia.googlecode.com/svn/trunk@14655 2bbb7eff-a529-9590-31e7-b0007b416f81
|
|
|
|
|
|
|
|
|
|
|
| |
BUG=skia:2378
R=fmalita@chromium.org, mtklein@google.com
Author: mtklein@chromium.org
Review URL: https://codereview.chromium.org/277613002
git-svn-id: http://skia.googlecode.com/svn/trunk@14653 2bbb7eff-a529-9590-31e7-b0007b416f81
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
The order of arguments in these structs is arbitrary, so we might as well arrange them to optimize something. Putting the paints at the front means the logic to find the paint is a lot more concise: it's usually just ptr+0, or *(ptr+0) when the SkPaint is optional.
This considerably reduces the size of the jump table in IsDraw::operator().
BUG=skia:2378
R=fmalita@chromium.org, mtklein@google.com
Author: mtklein@chromium.org
Review URL: https://codereview.chromium.org/272673002
git-svn-id: http://skia.googlecode.com/svn/trunk@14634 2bbb7eff-a529-9590-31e7-b0007b416f81
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
- Allow any return type from SkRecord mutators and visitors;
- update existing calls to mutate and visit;
- convert match to operator() in SkRecordPattern;
- tidy up a few inelegant bits of old code in tests.
The net result is that the generated code is much clearer. All the mutate() calls
inline as you'd hope, and you can now actually follow along with the disassembly.
BUG=skia:2378
R=fmalita@chromium.org, bungeman@google.com, mtklein@google.com
Author: mtklein@chromium.org
Review URL: https://codereview.chromium.org/273643007
git-svn-id: http://skia.googlecode.com/svn/trunk@14631 2bbb7eff-a529-9590-31e7-b0007b416f81
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
This way GCC/Clang don't generate the magic static code to call the
pointless T::T() once in a threadsafe way. = {} is plenty initialized.
BUG=skia:2378
Committed: http://code.google.com/p/skia/source/detail?r=14623
R=bungeman@google.com, mtklein@google.com
Author: mtklein@chromium.org
Review URL: https://codereview.chromium.org/270353003
git-svn-id: http://skia.googlecode.com/svn/trunk@14626 2bbb7eff-a529-9590-31e7-b0007b416f81
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
(https://codereview.chromium.org/270353003/)
Reason for revert:
breaks windows bots
Original issue's description:
> Statically initialize those zero-size singletons.
>
> This way GCC/Clang don't generate the magic static code to call the
> pointless T::T() once in a threadsafe way. = {} is plenty initialized.
>
> BUG=skia:2378
>
> Committed: http://code.google.com/p/skia/source/detail?r=14623
R=bungeman@google.com, mtklein@chromium.org
TBR=bungeman@google.com, mtklein@chromium.org
NOTREECHECKS=true
NOTRY=true
BUG=skia:2378
Author: mtklein@google.com
Review URL: https://codereview.chromium.org/270363005
git-svn-id: http://skia.googlecode.com/svn/trunk@14625 2bbb7eff-a529-9590-31e7-b0007b416f81
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
This way GCC/Clang don't generate the magic static code to call the
pointless T::T() once in a threadsafe way. = {} is plenty initialized.
BUG=skia:2378
R=bungeman@google.com, mtklein@google.com
Author: mtklein@chromium.org
Review URL: https://codereview.chromium.org/270353003
git-svn-id: http://skia.googlecode.com/svn/trunk@14623 2bbb7eff-a529-9590-31e7-b0007b416f81
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
This is particularly helpful for SkRecord::replace<NoOp>, which now doesn't go
off and allocate a pointless byte.
BUG=skia:2378
R=fmalita@chromium.org, bungeman@google.com, mtklein@google.com
Author: mtklein@chromium.org
Review URL: https://codereview.chromium.org/269543025
git-svn-id: http://skia.googlecode.com/svn/trunk@14622 2bbb7eff-a529-9590-31e7-b0007b416f81
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
save-drawA-drawB-drawC-restore always means drawA-drawB-drawC,
no matter what flags we use for save().
This one triggers all over the silk SKPs and in several of the bot SKPs, typically as save-drawBitmap-restore.
BUG=skia:2378
R=robertphillips@google.com, mtklein@google.com
Author: mtklein@chromium.org
Review URL: https://codereview.chromium.org/267793006
git-svn-id: http://skia.googlecode.com/svn/trunk@14618 2bbb7eff-a529-9590-31e7-b0007b416f81
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
This is like SkPictureRecord's remove_save_layer1 but works with all draw calls.
Interesting patterns removed:
SaveLayer-DrawRect-Restore: Silk SKPs, desk_weather
SaveLayer-DrawPath-Restore: desk_carsvg, desk_wowwiki, tabl_androidpolice
SaveLayer-DrawPosTextH-Restore: tabl_android_police
There may be others, but I stopped looking.
BUG=skia:2378
R=robertphillips@google.com, mtklein@google.com
Author: mtklein@chromium.org
Review URL: https://codereview.chromium.org/269813010
git-svn-id: http://skia.googlecode.com/svn/trunk@14610 2bbb7eff-a529-9590-31e7-b0007b416f81
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Will keep thinking about the best way to handle this:
- leave as-is
- tag the records
- some range check on T::kType
- just list all Draw* in IsDraw
BUG=skia:2378
R=fmalita@chromium.org, mtklein@google.com
Author: mtklein@chromium.org
Review URL: https://codereview.chromium.org/269543023
git-svn-id: http://skia.googlecode.com/svn/trunk@14602 2bbb7eff-a529-9590-31e7-b0007b416f81
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
This is a mid-level library for finding patterns of commands in an SkRecord. At the API level, it's a bit regex inspired. Some examples:
- Pattern1<Is<DrawRect>> matches a single DrawRect
- Pattern1<Star<Is<DrawRect>>> matches 0 or more DrawRects
- Pattern2<Is<ClipRect>, Is<DrawRect>> matches a single clip rect followed by a single draw rect
- Pattern3<Is<Save>, Star<IsDraw>, Is<Restore>> matches a single Save, followed by any number of Draws, followed by Restore
- Pattern1<Or<Is<DrawRect>, Is<ClipRect>>> matches a DrawRect or a ClipRect
- Pattern1<Not<Is<ClipRect>>> matches a command that's notClipRect.
Once you have a pattern, you can call .search() on it to step through ranges of matching commands. This means patterns can replace most of the custom iteration logic for optimization passes: the generic pattern searching steps through all the optimization candidates, which optimization-specific code further inspects and mutates.
SkRecordTraits is now unused. Bye bye!
Generated code and performance of SkRecordOpts is very similar to what it was before. (I had to use SK_ALWAYS_INLINE in a few places to make this so.)
BUG=skia:2378
R=fmalita@chromium.org, bungeman@google.com, mtklein@google.com
Author: mtklein@chromium.org
Review URL: https://codereview.chromium.org/263063002
git-svn-id: http://skia.googlecode.com/svn/trunk@14582 2bbb7eff-a529-9590-31e7-b0007b416f81
|
|
|
|
|
|
|
|
|
|
|
| |
BUG=skia:2378
R=fmalita@chromium.org, mtklein@google.com
Author: mtklein@chromium.org
Review URL: https://codereview.chromium.org/251133008
git-svn-id: http://skia.googlecode.com/svn/trunk@14455 2bbb7eff-a529-9590-31e7-b0007b416f81
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
On Mike's suggestion, I tested out not doing any empty-clip check at all in
SkRecordDraw, given that mostly we'll do that again anyway inside SkCanvas.
Most SKPs are identical to the status quo, whether bot or silk, played back in tiles
or full. Average playback performance, both arithmetic and geometric mean, is also
unchanged.
A handful of SKPs do draw faster or slower reliably, particularly when tiled. E.g. a
cnn tile draws about 40% faster, a cuteoverload tile about 20% slower. Their profiles
look pretty much the same before and after, so I can't really explain the changes.
I'd say, given that performance is mostly identical and very identical in bulk,
we might as well remove this code. It's nice to keep SkRecordDraw as dumb as possible.
BUG=skia:2378
R=reed@google.com, fmalita@chromium.org, mtklein@google.com, borenet@google.com
Author: mtklein@chromium.org
Review URL: https://codereview.chromium.org/258183002
git-svn-id: http://skia.googlecode.com/svn/trunk@14433 2bbb7eff-a529-9590-31e7-b0007b416f81
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
No measurable playback speed difference, but simpler code.
This makes sense: I'm seeing SkRecordDraw at ~0.25% of playback cost. We can
pretty much do whatever we want in there for free if it helps avoid real work.
BUG=skia:2378
R=fmalita@chromium.org, mtklein@google.com, reed@google.com
Author: mtklein@chromium.org
Review URL: https://codereview.chromium.org/260143002
git-svn-id: http://skia.googlecode.com/svn/trunk@14419 2bbb7eff-a529-9590-31e7-b0007b416f81
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Simplified skip logic by always running clip commands. No performance difference on bot or silk SKPs.
BUG=skia:2378
R=bungeman@google.com, fmalita@chromium.org, mtklein@google.com
Author: mtklein@chromium.org
Review URL: https://codereview.chromium.org/258693006
git-svn-id: http://skia.googlecode.com/svn/trunk@14410 2bbb7eff-a529-9590-31e7-b0007b416f81
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
It used to be an unenforced requirement that callers take ownership of
the command which was replaced when calling SkRecord::replace. Now we
can enforce it, by splitting replace into two modes:
- T* replace(i): always destroys the existing command for you
- T* replace(i, proofOfAdoption): proofOfAdoption is checked to make
sure the caller has adopted the existing command before replacing it.
BUG=skia:2378
R=fmalita@chromium.org, mtklein@google.com
Author: mtklein@chromium.org
Review URL: https://codereview.chromium.org/248053008
git-svn-id: http://skia.googlecode.com/svn/trunk@14352 2bbb7eff-a529-9590-31e7-b0007b416f81
|
|
|
|
|
|
|
|
|
|
|
| |
BUG=skia:2378
R=reed@google.com, danakj@chromium.org, enne@chromium.org, mtklein@google.com, robertphillips@google.com
Author: mtklein@chromium.org
Review URL: https://codereview.chromium.org/248033002
git-svn-id: http://skia.googlecode.com/svn/trunk@14351 2bbb7eff-a529-9590-31e7-b0007b416f81
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
(https://codereview.chromium.org/243853006/)"
associated chrome change (to be committed with DEPS roll)
https://codereview.chromium.org/248693002/
This reverts commit bfaceb53f58c9625b5471fcff35b5ca9ca3ae29c.
TBR=bsalomon@google.com
Review URL: https://codereview.chromium.org/248083002
git-svn-id: http://skia.googlecode.com/svn/trunk@14321 2bbb7eff-a529-9590-31e7-b0007b416f81
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
(https://codereview.chromium.org/243853006/)
Reason for revert:
need to test more the code-path where we don't opt into the new virtuals
Original issue's description:
> make drawText calls non-virtual, to ease SkFont and TextBlob
>
> BUG=skia:
>
> Committed: http://code.google.com/p/skia/source/detail?r=14307
R=robertphillips@google.com
TBR=robertphillips@google.com
NOTREECHECKS=true
NOTRY=true
BUG=skia:
Author: reed@google.com
Review URL: https://codereview.chromium.org/247983003
git-svn-id: http://skia.googlecode.com/svn/trunk@14314 2bbb7eff-a529-9590-31e7-b0007b416f81
|
|
|
|
|
|
|
|
|
|
|
| |
BUG=skia:
R=robertphillips@google.com
Author: reed@google.com
Review URL: https://codereview.chromium.org/243853006
git-svn-id: http://skia.googlecode.com/svn/trunk@14307 2bbb7eff-a529-9590-31e7-b0007b416f81
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Also, make sure to undef macros after we're done with them.
BUG=skia:
NOTRY=true
R=bungeman@google.com, mtklein@google.com
Author: mtklein@chromium.org
Review URL: https://codereview.chromium.org/247573003
git-svn-id: http://skia.googlecode.com/svn/trunk@14305 2bbb7eff-a529-9590-31e7-b0007b416f81
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
This adds back two optimizations from SkPicture: drawPosText strength reduction to drawPosTextH, and pointless save-foo-restore blocks are noop'd away.
The small-T optimization in SkRecord gets in the way of implementing replace(), so I removed it.
Just to keep the API focused, I removed the methods on SkRecord that iterate over i for you; it's just as efficient to do it yourself, and all of the interesting code does its own custom iteration.
BUG=skia:2378
R=fmalita@chromium.org, mtklein@google.com
Author: mtklein@chromium.org
Review URL: https://codereview.chromium.org/245853002
git-svn-id: http://skia.googlecode.com/svn/trunk@14300 2bbb7eff-a529-9590-31e7-b0007b416f81
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
also, call the new SkRecordOptimize in bench_playback
BUG=skia:2378
R=fmalita@chromium.org, mtklein@google.com
Author: mtklein@chromium.org
Review URL: https://codereview.chromium.org/243243003
git-svn-id: http://skia.googlecode.com/svn/trunk@14277 2bbb7eff-a529-9590-31e7-b0007b416f81
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Optional arguments to SkCanvas calls leaked refs (but not memory) because we
didn't destruct the optional objects (really, just SkPaint: other optional args
are all POD). This adds Optional and PODArray, where Optional makes sure to
call the destructor.
I overlooked that SkPictureRecord really does call paint.computeFastBounds().
Do the same in SkRecordDraw.
BUG=skia:2378
R=reed@google.com, mtklein@google.com, tomhudson@google.com
Author: mtklein@chromium.org
Review URL: https://codereview.chromium.org/235983015
git-svn-id: http://skia.googlecode.com/svn/trunk@14197 2bbb7eff-a529-9590-31e7-b0007b416f81
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
PS 1) factor apart into canSkip, draw, updateClip
PS 2) use quickRejectY for text
BUG=skia:2378
R=fmalita@chromium.org, reed@google.com, mtklein@google.com, tomhudson@google.com
Author: mtklein@chromium.org
Review URL: https://codereview.chromium.org/233493004
git-svn-id: http://skia.googlecode.com/svn/trunk@14192 2bbb7eff-a529-9590-31e7-b0007b416f81
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Also, switch Skia internal tools over to use the public headers where
possible. Where it's not, the tools -Isrc/record explicitly now, and
if it's not obvious, note why they don't use SkRecording.h.
BUG=skia:2378
R=reed@google.com, mtklein@google.com, fmalita@chromium.org
Author: mtklein@chromium.org
Review URL: https://codereview.chromium.org/231853006
git-svn-id: http://skia.googlecode.com/svn/trunk@14191 2bbb7eff-a529-9590-31e7-b0007b416f81
|
|
|
|
|
|
|
|
|
|
|
| |
BUG=skia:
R=reed@google.com, mtklein@google.com
Author: mtklein@chromium.org
Review URL: https://codereview.chromium.org/235253002
git-svn-id: http://skia.googlecode.com/svn/trunk@14158 2bbb7eff-a529-9590-31e7-b0007b416f81
|
|
|
|
|
|
|
|
|
|
|
| |
BUG=skia:2378
R=reed@google.com, robertphillips@google.com, mtklein@google.com, fmalita@chromium.org
Author: mtklein@chromium.org
Review URL: https://codereview.chromium.org/233053005
git-svn-id: http://skia.googlecode.com/svn/trunk@14146 2bbb7eff-a529-9590-31e7-b0007b416f81
|
|
|
|
|
|
|
|
|
|
|
| |
BUG=skia:2378
R=fmalita@chromium.org, mtklein@google.com, fmalita@google.com
Author: mtklein@chromium.org
Review URL: https://codereview.chromium.org/231933003
git-svn-id: http://skia.googlecode.com/svn/trunk@14126 2bbb7eff-a529-9590-31e7-b0007b416f81
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
- Adds tests for SkRecordDraw's two main features: cull- and clip- based skipping.
- Adds full SkCanvas semantic mode to SkRecorder, so it can be used as a target for SkRecordDraw.
BUG=skia:2378
R=fmalita@chromium.org, mtklein@google.com
Author: mtklein@chromium.org
Review URL: https://codereview.chromium.org/231653002
git-svn-id: http://skia.googlecode.com/svn/trunk@14124 2bbb7eff-a529-9590-31e7-b0007b416f81
|
|
|
|
| |
git-svn-id: http://skia.googlecode.com/svn/trunk@14102 2bbb7eff-a529-9590-31e7-b0007b416f81
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
- Allow stateful functors; allow visit()/mutate() at a given index; add count().
- Annotate cull push/pop pairs on the PushCull records. (tested)
- Use those annotations to skip ahead in SkRecordDraw. (not yet tested beyond dm --skr)
- Make SkRecordDraw a function, move its implementation to a .cpp.
BUG=skia:2378
R=fmalita@chromium.org, mtklein@google.com
Author: mtklein@chromium.org
Review URL: https://codereview.chromium.org/229523002
git-svn-id: http://skia.googlecode.com/svn/trunk@14101 2bbb7eff-a529-9590-31e7-b0007b416f81
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
- add SK_OVERRIDE for SkCanvas methods in SkRecorder
- start on unit tests, here just for SkRecord itself
BUG=skia:2378
R=fmalita@google.com, mtklein@google.com, fmalita@chromium.org
Author: mtklein@chromium.org
Review URL: https://codereview.chromium.org/228723003
git-svn-id: http://skia.googlecode.com/svn/trunk@14097 2bbb7eff-a529-9590-31e7-b0007b416f81
|
|
|
|
|
|
|
|
|
|
|
| |
BUG=skia:2378
R=fmalita@google.com, mtklein@google.com, fmalita@chromium.org
Author: mtklein@chromium.org
Review URL: https://codereview.chromium.org/224723026
git-svn-id: http://skia.googlecode.com/svn/trunk@14091 2bbb7eff-a529-9590-31e7-b0007b416f81
|
|
Record performance as measured by bench_record (out/Release/bench_record --skr) improves by at least 1.9x, at most 6.7x, arithmetic mean 2.6x, geometric mean 3.0x. So, good.
Correctness as measured by DM (out/Debug/dm --skr) is ~ok. One GM (shadertext2) fails because we're assuming all paint effects are immutable, but SkShaders are still mutable.
To do after this CL:
- measure playback speed
- catch up feature-wise to SkPicture
- match today's playback speed
BUG=skia:
R=robertphillips@google.com, bsalomon@google.com, reed@google.com, mtklein@google.com
Author: mtklein@chromium.org
Review URL: https://codereview.chromium.org/206313003
git-svn-id: http://skia.googlecode.com/svn/trunk@14010 2bbb7eff-a529-9590-31e7-b0007b416f81
|