- Remove one especially chatty event, and one pointless test
- Use TRACE_FUNC everywhere, rather than manual strings
Bug: skia:
Change-Id: Icb795294009150ca9a260436738d79546a733337
Reviewed-on: https://skia-review.googlesource.com/26701
Reviewed-by: Mike Klein <mtklein@chromium.org>
Commit-Queue: Brian Osman <brianosman@google.com>
I was going to be clever here and only clamp when we
know the inputs are out of range, but this filter is
rare and slow enough that I think I'd rather it just
be super paranoid safe.
The test crashes without this fix and passes with it.
Change-Id: I4e17aad2b5c1e96180ce8d73b97bee746cf985c2
Reviewed-on: https://skia-review.googlesource.com/26702
Commit-Queue: Brian Osman <brianosman@google.com>
Reviewed-by: Brian Osman <brianosman@google.com>
DAA is:
1. Much simpler than AAA.
SkScan_AAAPath.cpp is about 1700 lines.
SkScan_DAAPath.cpp is about 300 lines.
The whole DAA CL is only about 800 lines.
2. Much faster than AAA for complicated paths.
The speedup applies to GL backend (including ccpr)!
Here's the frame time of 'SampleApp --slide Chart' on macbook pro:
AAA-raster: 33ms
DAA-raster: 21ms
AAA-gl: 30ms
DAA-gl: 20ms
AAA-ccpr: 18ms
DAA-ccpr: 12ms
My linux desktop doesn't have SSE3 so the speedup is smaller
(~25% for Chart). I believe that DAA is so fast that I can enable
it for any paths (AAA is not enabled by default for complicated
paths because it is slow; hence our older supersampling scan
converter is used for stroking on Chart for AAA-xxx config.)
3. The SkCoverageDelta is suitable for threaded backend with
out-of-order concurrent scan conversion as commented in the source
code. Maybe we can also just send deltas to GPU.
4. Similar to most analytic path renderers, the quality is on the best
ground-truth level, unless there are intersections within a pixel.
The intersections look good to my eyes although theoretically that
could be arbitrary far from the ground truth (see my AAA slides).
5. For simple paths, such as circle, triangle, rrect, etc., DAA is
slower than AAA. But DAA is faster than our older supersampling
scan converter in most cases. As those simple paths usually don't
constitute the bottleneck of a picture (skp or svg), I strongly
recommend use DAA.
6. DAA also heavily favors blitMask so it may work quite well with
SkRasterPipeline and SkRasterPipelineBlitter.
Finally, please check https://skia-review.googlesource.com/c/22420/
which accelerate DAA by specializing blitCoverageDeltas for
SkARGB32_Blitter and SkARGB32_Black_Blitter. It brings a little(<5%)
speedup. But I couldn't figure out how to reduce the duplicate code
so I don't intend to land it.
Bug: skia:
Change-Id: I3b7ed6a727447922e645b1acb737a506e7c09a4c
Reviewed-on: https://skia-review.googlesource.com/19666
Reviewed-by: Mike Reed <reed@google.com>
Reviewed-by: Cary Clark <caryclark@google.com>
Commit-Queue: Yuqian Li <liyuqian@google.com>
Our new srgb_color_filter (in gm/srgb_color_filter.cpp) is awfully
similar to the existing srgb_colorfilter (in gm/srgb.cpp).
Let's rename the new one.
Change-Id: I8c7816c377cc4342388be51632353882a1bc5241
Reviewed-on: https://skia-review.googlesource.com/26525
Reviewed-by: Mike Reed <reed@google.com>
Commit-Queue: Mike Klein <mtklein@chromium.org>
This reverts commit fa3ed03720.
Reason for revert: GM is failing on a lot of bots
Original change's description:
> Add GM to test SkToSRGBColorFilter
>
> Bug: skia:
> Change-Id: If342ad5503d1b427f2d04ce15b75f0f7fa2706c1
> Reviewed-on: https://skia-review.googlesource.com/26426
> Reviewed-by: Mike Klein <mtklein@google.com>
> Commit-Queue: Brian Osman <brianosman@google.com>
TBR=djsollen@google.com,mtklein@google.com,brianosman@google.com
Change-Id: I92fe179ead6d115b32b3a9533b8ee0e10bb3ee43
No-Presubmit: true
No-Tree-Checks: true
No-Try: true
Bug: skia:
Reviewed-on: https://skia-review.googlesource.com/26522
Reviewed-by: Kevin Lubick <kjlubick@google.com>
Commit-Queue: Kevin Lubick <kjlubick@google.com>
Bug: skia:
Change-Id: If342ad5503d1b427f2d04ce15b75f0f7fa2706c1
Reviewed-on: https://skia-review.googlesource.com/26426
Reviewed-by: Mike Klein <mtklein@google.com>
Commit-Queue: Brian Osman <brianosman@google.com>
Bug: skia:
Change-Id: I9f04b8ce778349218ccd55673bdd1d16a192383c
Reviewed-on: https://skia-review.googlesource.com/26422
Reviewed-by: Mike Klein <mtklein@chromium.org>
Commit-Queue: Mike Klein <mtklein@chromium.org>
This was used by subclasses of GrLegacyMeshDrawOp.
Change-Id: I421589dfdc253bdf43aeac9fa9af0647c70811dd
Reviewed-on: https://skia-review.googlesource.com/25804
Reviewed-by: Robert Phillips <robertphillips@google.com>
Commit-Queue: Brian Salomon <bsalomon@google.com>
This is mostly dead code.
In order to make it truly dead, we need to opt drawing unpremul images
into SkRasterPipelineBlitter. They had been handled by
SkLinearBitmapPipeline, but can't be draw by SkBitmapProcLegacyShader.
Drawing unpremul images is tested by the GM all_variants_8888, which
gave us trouble last time around (serialize-8888 drew right, 8888 wrong)
but now draws fine. I think this was probably also the root of the
revert, drawing some unpremul image in Chrome's tests somewhere.
Change-Id: I453f9df44ade807316935921cbae82961e2f08aa
Reviewed-on: https://skia-review.googlesource.com/24862
Reviewed-by: Florin Malita <fmalita@chromium.org>
Commit-Queue: Mike Klein <mtklein@chromium.org>
CQ_INCLUDE_TRYBOTS=skia.primary:Test-Debian9-GCC-GCE-CPU-AVX2-x86_64-Release-SKNX_NO_SIMD
Bug: skia:
Change-Id: If6f0d0a57463bf99a66d674e65a62ce3931d0116
Reviewed-on: https://skia-review.googlesource.com/24644
Commit-Queue: Mike Reed <reed@google.com>
Reviewed-by: Mike Klein <mtklein@chromium.org>
Bug: skia:
Change-Id: I97af420d1c3270e24e5d0959237b8163faa9e069
Reviewed-on: https://skia-review.googlesource.com/24646
Reviewed-by: Brian Osman <brianosman@google.com>
Commit-Queue: Ethan Nicholas <ethannicholas@google.com>
Change-Id: Iaecae142dcf268517f8918d9f4fa6a64db194fab
Reviewed-on: https://skia-review.googlesource.com/23942
Commit-Queue: Mike Klein <mtklein@google.com>
Reviewed-by: Ben Wagner <bungeman@google.com>
Initial implementation of a GPU path renderer that draws antialiased
paths by counting coverage in an offscreen buffer.
Initially disabled until it has had time to soak.
Bug: skia:
Change-Id: I003d8cfdf8dc62641581b5ea2dc4f0aa00108df6
Reviewed-on: https://skia-review.googlesource.com/21541
Commit-Queue: Chris Dalton <csmartdalton@google.com>
Reviewed-by: Greg Daniel <egdaniel@google.com>
Reviewed-by: Brian Salomon <bsalomon@google.com>
Reviewed-by: Robert Phillips <robertphillips@google.com>
Bug: skia:
Change-Id: I028c85f692b2e54e64be301940708387014ee0c7
Reviewed-on: https://skia-review.googlesource.com/23540
Reviewed-by: Jim Van Verth <jvanverth@google.com>
Commit-Queue: Ethan Nicholas <ethannicholas@google.com>
Bug: skia:
Change-Id: Id089b9ead7a21e903b001006dffff2381efd4ba3
Reviewed-on: https://skia-review.googlesource.com/23582
Reviewed-by: Brian Osman <brianosman@google.com>
Commit-Queue: Ethan Nicholas <ethannicholas@google.com>
Bug: skia:
Change-Id: I93d117c22ae2b374294f6a5e961c497ac2c92b09
Reviewed-on: https://skia-review.googlesource.com/23301
Reviewed-by: Brian Osman <brianosman@google.com>
Commit-Queue: Ethan Nicholas <ethannicholas@google.com>
This reverts commit 27b3d272a8.
Reason for revert: guard has landed in android
Original change's description:
> Revert "Experimental blur code with 32 bit fix."
>
> This reverts commit d4b2c537d0.
>
> Reason for revert: speculative fix for android-roll
>
> java.lang.AssertionError: expected:<0> but was:<255>
> at org.junit.Assert.fail(Assert.java:88)
> at org.junit.Assert.failNotEquals(Assert.java:834)
> at org.junit.Assert.assertEquals(Assert.java:645)
> at org.junit.Assert.assertEquals(Assert.java:631)
> at android.graphics.cts.BlurMaskFilterTest.verifyColor(BlurMaskFilterTest.java:79)
> at android.graphics.cts.BlurMaskFilterTest.verifyQuadrants(BlurMaskFilterTest.java:72)
> at android.graphics.cts.BlurMaskFilterTest.testBlurMaskFilter(BlurMaskFilterTest.java:56)
>
> Original change's description:
> > Experimental blur code with 32 bit fix.
> >
> > This uses a new method of blurring that runs the three
> > passes of the box filter in a single pass. This implementation
> > currently only does 1x1 pixel at a time, but it should be simple
> > to expand to 4x4 pixels at a time.
> >
> > On the blur_10_normal_high_quality benchmark, the new is 7% faster
> > than the old code. For the blur_100.50_normal_high_quality
> > benchmark, the new code is 11% slower.
> >
> > Bug: skia:
> > Change-Id: I847270906b0ceac1dfbf43ab5446756689ef660f
> > Reviewed-on: https://skia-review.googlesource.com/22700
> > Reviewed-by: Mike Reed <reed@google.com>
> > Commit-Queue: Herb Derby <herb@google.com>
>
> TBR=herb@google.com,reed@google.com
>
> Change-Id: Ie84f6bf8872cae08c06d679f0c2f2e6c3d8a02a2
> No-Presubmit: true
> No-Tree-Checks: true
> No-Try: true
> Bug: skia:
> Reviewed-on: https://skia-review.googlesource.com/22880
> Reviewed-by: Mike Reed <reed@google.com>
> Commit-Queue: Mike Reed <reed@google.com>
TBR=herb@google.com,reed@google.com
Change-Id: I393d1c05f83ccf98137201bc7b4e7d8e8b0e8742
No-Presubmit: true
No-Tree-Checks: true
No-Try: true
Bug: skia:
Reviewed-on: https://skia-review.googlesource.com/23121
Reviewed-by: Mike Reed <reed@google.com>
Commit-Queue: Mike Reed <reed@google.com>
This reverts commit d4b2c537d0.
Reason for revert: speculative fix for android-roll
java.lang.AssertionError: expected:<0> but was:<255>
at org.junit.Assert.fail(Assert.java:88)
at org.junit.Assert.failNotEquals(Assert.java:834)
at org.junit.Assert.assertEquals(Assert.java:645)
at org.junit.Assert.assertEquals(Assert.java:631)
at android.graphics.cts.BlurMaskFilterTest.verifyColor(BlurMaskFilterTest.java:79)
at android.graphics.cts.BlurMaskFilterTest.verifyQuadrants(BlurMaskFilterTest.java:72)
at android.graphics.cts.BlurMaskFilterTest.testBlurMaskFilter(BlurMaskFilterTest.java:56)
Original change's description:
> Experimental blur code with 32 bit fix.
>
> This uses a new method of blurring that runs the three
> passes of the box filter in a single pass. This implementation
> currently only does 1x1 pixel at a time, but it should be simple
> to expand to 4x4 pixels at a time.
>
> On the blur_10_normal_high_quality benchmark, the new is 7% faster
> than the old code. For the blur_100.50_normal_high_quality
> benchmark, the new code is 11% slower.
>
> Bug: skia:
> Change-Id: I847270906b0ceac1dfbf43ab5446756689ef660f
> Reviewed-on: https://skia-review.googlesource.com/22700
> Reviewed-by: Mike Reed <reed@google.com>
> Commit-Queue: Herb Derby <herb@google.com>
TBR=herb@google.com,reed@google.com
Change-Id: Ie84f6bf8872cae08c06d679f0c2f2e6c3d8a02a2
No-Presubmit: true
No-Tree-Checks: true
No-Try: true
Bug: skia:
Reviewed-on: https://skia-review.googlesource.com/22880
Reviewed-by: Mike Reed <reed@google.com>
Commit-Queue: Mike Reed <reed@google.com>
This is mainly for getting ready to start adding lots of metal backend code.
I've also update the "gpu tools" target to require ARC with involved updating
one IOS file in there.
Bug: skia:
Change-Id: Ied22e8fe7532445cc274efb529e3450654a6614b
Reviewed-on: https://skia-review.googlesource.com/22484
Commit-Queue: Greg Daniel <egdaniel@google.com>
Reviewed-by: Brian Salomon <bsalomon@google.com>
Reviewed-by: Mike Klein <mtklein@google.com>
- SkImageSubset becomes SkKeyedImage
- SkPDFCanvas::onDraw{Bitmap, Image} go away
- Remove SkPDFCanvas: base classes now do the right thing.
- SkPDFDevice::draw{Bitmap,Image}{Rect,}() simplified
- 244 fewer SLOC.
All but a few PDFs are identical, those rasterize almost the same.
Change-Id: I3ceb3b8935c689719cedf1ad544b0407b5c1733e
Reviewed-on: https://skia-review.googlesource.com/22218
Commit-Queue: Hal Canary <halcanary@google.com>
Reviewed-by: Ben Wagner <bungeman@google.com>
This uses a new method of blurring that runs the three
passes of the box filter in a single pass. This implementation
currently only does 1x1 pixel at a time, but it should be simple
to expand to 4x4 pixels at a time.
On the blur_10_normal_high_quality benchmark, the new is 7% faster
than the old code. For the blur_100.50_normal_high_quality
benchmark, the new code is 11% slower.
Bug: skia:
Change-Id: I847270906b0ceac1dfbf43ab5446756689ef660f
Reviewed-on: https://skia-review.googlesource.com/22700
Reviewed-by: Mike Reed <reed@google.com>
Commit-Queue: Herb Derby <herb@google.com>
This reverts commit d96ed9d0de.
Reason for revert: dm crashing
https://luci-milo.appspot.com/swarming/task/374d82c1d1263910/steps/symbolized_dm/0/stdout
Likely culprit:
unit test BlurDrawing
Stack trace:
/mnt/pd0/s/w/ir/out/Debug/dm(+0x2440eb) [0x568770eb]
linux-gate.so.1(__kernel_sigreturn+0) [0xf7714ca0]
/mnt/pd0/s/w/ir/out/Debug/dm(_ZNK16SkMaskBlurFilter11blurOneScanENS_10FilterInfoEPKhjS2_PhjS3_+0x236) [0x5766301e]
/mnt/pd0/s/w/ir/out/Debug/dm(_ZNK16SkMaskBlurFilter4blurERK6SkMaskPS0_+0x285) [0x57663491]
/mnt/pd0/s/w/ir/out/Debug/dm(_ZN10SkBlurMask7BoxBlurEP6SkMaskRKS0_f11SkBlurStyle13SkBlurQualityP8SkIPointb+0x5c) [0x5720d48e]
/mnt/pd0/s/w/ir/out/Debug/dm(_ZNK20SkBlurMaskFilterImpl10filterMaskEP6SkMaskRKS0_RK8SkMatrixP8SkIPoint+0x67) [0x5720e427]
/mnt/pd0/s/w/ir/out/Debug/dm(_ZNK20SkBlurMaskFilterImpl17filterRectsToNineEPK6SkRectiRK8SkMatrixRK7SkIRectPN12SkMaskFilter9NinePatchE+0x579) [0x5721a247]
/mnt/pd0/s/w/ir/out/Debug/dm(_ZNK12SkMaskFilter10filterPathERK6SkPathRK8SkMatrixRK12SkRasterClipP9SkBlitterN11SkStrokeRec9InitStyleE+0xf6) [0x5706c2bc]
/mnt/pd0/s/w/ir/out/Debug/dm(_ZNK6SkDraw11drawDevPathERK6SkPathRK7SkPaintbP9SkBlitterb+0x1bb) [0x57034f1b]
/mnt/pd0/s/w/ir/out/Debug/dm(_ZNK6SkDraw8drawPathERK6SkPathRK7SkPaintPK8SkMatrixbbP9SkBlitter+0x47f) [0x57035dd3]
/mnt/pd0/s/w/ir/out/Debug/dm(_ZN14SkBitmapDevice8drawPathERK6SkPathRK7SkPaintPK8SkMatrixb+0x6e) [0x56f937fc]
/mnt/pd0/s/w/ir/out/Debug/dm(_ZN8SkCanvas10onDrawPathERK6SkPathRK7SkPaint+0x39d) [0x56fb12e5]
/mnt/pd0/s/w/ir/out/Debug/dm(_ZN8SkCanvas8drawPathERK6SkPathRK7SkPaint+0x18) [0x56fad926]
/mnt/pd0/s/w/ir/out/Debug/dm(+0x3f71b3) [0x56a2a1b3]
/mnt/pd0/s/w/ir/out/Debug/dm(+0x2450bd) [0x568780bd]
/mnt/pd0/s/w/ir/out/Debug/dm(+0x2450fb) [0x568780fb]
/mnt/pd0/s/w/ir/out/Debug/dm(+0xae53e7) [0x571183e7]
/mnt/pd0/s/w/ir/out/Debug/dm(_ZNKSt8functionIFvvEEclEv+0x20) [0x569f98de]
/mnt/pd0/s/w/ir/out/Debug/dm(_ZN12SkThreadPool4LoopEPv+0x298) [0x57045b2f]
/mnt/pd0/s/w/ir/out/Debug/dm(+0xbbd92c) [0x571f092c]
/lib/i386-linux-gnu/libpthread.so.0(+0x627a) [0xf76e827a]
/lib/i386-linux-gnu/libc.so.6(clone+0x66) [0xf70ecb56]
Segmentation fault
Original change's description:
> Experimental blur code.
>
> This uses a new method of blurring that runs the three
> passes of the box filter in a single pass. This implementation
> currently only does 1x1 pixel at a time, but it should be simple
> to expand to 4x4 pixels at a time.
>
> On the blur_10_normal_high_quality benchmark, the new is 7% faster
> than the old code. For the blur_100.50_normal_high_quality
> benchmark, the new code is 11% slower.
>
> Change-Id: Iea37294abc7c27de5ad569adf8bc62df77eafd02
> Reviewed-on: https://skia-review.googlesource.com/21739
> Commit-Queue: Herb Derby <herb@google.com>
> Reviewed-by: Mike Reed <reed@google.com>
TBR=herb@google.com,reed@google.com
Change-Id: I9e896c548d0a4cd3308d6a311c8bd16719a08a85
No-Presubmit: true
No-Tree-Checks: true
No-Try: true
Reviewed-on: https://skia-review.googlesource.com/22421
Reviewed-by: Florin Malita <fmalita@google.com>
Commit-Queue: Florin Malita <fmalita@google.com>
This uses a new method of blurring that runs the three
passes of the box filter in a single pass. This implementation
currently only does 1x1 pixel at a time, but it should be simple
to expand to 4x4 pixels at a time.
On the blur_10_normal_high_quality benchmark, the new is 7% faster
than the old code. For the blur_100.50_normal_high_quality
benchmark, the new code is 11% slower.
Change-Id: Iea37294abc7c27de5ad569adf8bc62df77eafd02
Reviewed-on: https://skia-review.googlesource.com/21739
Commit-Queue: Herb Derby <herb@google.com>
Reviewed-by: Mike Reed <reed@google.com>
Remove #defines that lived in gn_to_bp.py and
android_framework_defines.gni. These have been moved into a new file
in Android, SkUserConfigManual.h, in https://googleplex-android-review.git.corp.google.com/#/c/2519600/
Update gn_to_bp.py to include SkUserConfigManual.h, so it will still
result in using the same #defines.
Lately, we've found it difficult to guard changes behind a flag. e.g.
a change to drawing causes a CTS failure in Android, so we have to do
the following:
- put the change behind a flag, and add it to gn_to_bp.py or
android_framework_defines.gni
- generate new images on Android (by running CTS with external/skia
modified to not define the flag)
- create a CL in CTS that uses the new images
- land a CL in Skia that stops defining the flag
- when the Skia change lands, wait for the auto-roller to create a CL
that includes the change, stop the auto-roller, add the topic to the CTS
CL so the two can land at the same time
- land both Android changes (with TreeHugger)
- restart the Android auto-roller
With SkUserConfigManual.h (which lives in Android), the process will
be similar to Chromium:
- land a CL in Android's external/skia that defines a flag e.g.
SK_SUPPORT_LEGACY_FEATURE. Land without TreeHugger because it isn't used
in Skia and does not do anything
- land a change in Skia that changes behavior unless
SK_SUPPORT_LEGACY_FEATURE is defined. This will safely go through the
Android roll and not change any behavior for Android
- create two Android CLs - one in CTS to use the new images, and one in
external/skia to delete SK_SUPPORT_LEGACY_FEATURE. Set them to the same
topic and land them with TreeHugger
In the new process, there is no need to mess with the Android roll.
A downside to the new process is that we cannot test the android
framework defines without checking in to Android. But given how much
we've progressed in automating Android testing, this is fine.
Bug: b/63429612
Change-Id: Idfbaef2f4cae641a75fb6e7bf70428733a441336
Reviewed-on: https://skia-review.googlesource.com/22072
Commit-Queue: Leon Scroggins <scroggo@google.com>
Reviewed-by: Eric Boren <borenet@google.com>