Add fms op and instruction generation. Do fms and fnma
instruction selection.
TODO: Add the ops to Arm
Change-Id: I7e53abd7f4752eb99c31dcbff1f2ea7cf28af6c9
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/275197
Reviewed-by: Mike Klein <mtklein@google.com>
Commit-Queue: Mike Klein <mtklein@google.com>
Bug: skia:9984
Change-Id: Iaff520c6085303830f14bac4fe56830ba7360f64
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/275218
Reviewed-by: Brian Osman <brianosman@google.com>
Commit-Queue: Mike Reed <reed@google.com>
This is the easiest way to guarantee Op::fma_f32
actually fuses, by using platform intrinsics.
While implementing this we noticed that quad-pumping
was actually slower than double-pumping by about 25%,
and single-pumping was between the two. Switch from
quad to double pumping.
Change-Id: Ib93fd175fb8f6aaf49f769a95edfa9fd6b2674f6
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/275299
Commit-Queue: Mike Klein <mtklein@google.com>
Commit-Queue: Herb Derby <herb@google.com>
Reviewed-by: Herb Derby <herb@google.com>
Cluster edges were not enough.
Bug: skia:1003
Change-Id: Id2fdb7aa5dc2f6c4b03f1c841757796cf5c9b604
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/275220
Reviewed-by: Ben Wagner <bungeman@google.com>
Commit-Queue: Julia Lavrova <jlavrova@google.com>
Change-Id: I55e14b137d985fafd4e2c8fbc2cd76b27217699e
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/275217
Commit-Queue: Kevin Lubick <kjlubick@google.com>
Reviewed-by: Eric Boren <borenet@google.com>
This CL adds createProgramInfo and onPrePrepareDraws methods to:
NonAAStrokeRectOp
AAStrokeRectOp
NonAARectOp
Bug: skia:9455
Change-Id: Id7c0151d26c5daa390cd51b9c73c95f4f1fc0104
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/275039
Commit-Queue: Robert Phillips <robertphillips@google.com>
Reviewed-by: Greg Daniel <egdaniel@google.com>
9e9493f29d..c55fbc4064
git log 9e9493f29dbf..c55fbc4064aa --date=short --first-parent --format='%ad %ae %s'
2020-03-05 cnorthrop@google.com Capture/Replay: Fill out more ES 3.0 state in mid-execution capture
2020-03-05 courtneygo@google.com Vulkan: blacklist VK_FORMAT_R8G8B8_UNORM
2020-03-04 cnorthrop@google.com Capture/Replay: Add Sampler Object support to mid-execution capture
2020-03-04 ianelliott@google.com Vulkan: Suppress some flaky tests for Intel Linux driver
2020-03-04 cnorthrop@google.com Capture/Replay: Add Vertex Array Object support to mid-execution capture
2020-03-04 cnorthrop@google.com Capture/Replay: Add TransformFeedback Object support to mid-execution capture
2020-03-04 lexa.knyazev@gmail.com D3D11: Implement OES_draw_buffers_indexed
2020-03-04 ianelliott@google.com Vulkan: Disable forceOldRewriteStructSamplers for SwS-Android
2020-03-04 jmadill@chromium.org Fix up EGLSurfaceTest.
2020-03-04 xiaoxuan.liu@arm.com Vulkan: Re-enable RobustResourceInitTest.Texture test
2020-03-04 angle-autoroll@skia-public.iam.gserviceaccount.com Roll third_party/vulkan-tools/src a553ed74b7fe..1af57a3ec414 (1 commits)
2020-03-04 angle-autoroll@skia-public.iam.gserviceaccount.com Roll third_party/vulkan-validation-layers/src f32bf1c930c8..97484d689869 (16 commits)
2020-03-04 angle-autoroll@skia-public.iam.gserviceaccount.com Roll third_party/spirv-tools/src e1688b60caf7..a6d3a2dd41d4 (1 commits)
2020-03-04 angle-autoroll@skia-public.iam.gserviceaccount.com Roll third_party/SwiftShader 176f3a3dbbc9..126720bd2e57 (2 commits)
2020-03-04 angle-autoroll@skia-public.iam.gserviceaccount.com Roll third_party/glslang/src 56364b6b6026..8985fc91089f (2 commits)
2020-03-04 ancheng.qiao@arm.com Skip if GL_EXT_texture_compression_dxt1 isn't supported
Created with:
gclient setdep -r third_party/externals/angle2@c55fbc4064aa
If this roll has caused a breakage, revert this CL and stop the roller
using the controls here:
https://autoroll.skia.org/r/angle-skia-autoroll
Please CC herb@google.com on the revert to ensure that a human
is aware of the problem.
To report a problem with the AutoRoller itself, please file a bug:
https://bugs.chromium.org/p/skia/issues/entry?template=Autoroller+Bug
Documentation for the AutoRoller is here:
https://skia.googlesource.com/buildbot/+/master/autoroll/README.md
Cq-Include-Trybots: skia/skia.primary:Build-Debian9-Clang-x86_64-Release-ANGLE;skia/skia.primary:Test-Win10-Clang-AlphaR2-GPU-RadeonR9M470X-x86_64-Debug-All-ANGLE;skia/skia.primary:Test-Win10-Clang-Golo-GPU-QuadroP400-x86_64-Debug-All-ANGLE;skia/skia.primary:Test-Win10-Clang-NUC5i7RYH-GPU-IntelIris6100-x86_64-Debug-All-ANGLE;skia/skia.primary:Test-Win10-Clang-NUC6i5SYK-GPU-IntelIris540-x86_64-Debug-All-ANGLE;skia/skia.primary:Test-Win10-Clang-NUC8i5BEK-GPU-IntelIris655-x86_64-Debug-All-ANGLE;skia/skia.primary:Test-Win10-Clang-NUCD34010WYKH-GPU-IntelHD4400-x86_64-Debug-All-ANGLE
Bug: None
Tbr: herb@google.com
Change-Id: I58f5e0661a06553294ca3a65259b811d2ed22d6c
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/275211
Reviewed-by: skia-autoroll <skia-autoroll@skia-public.iam.gserviceaccount.com>
Commit-Queue: skia-autoroll <skia-autoroll@skia-public.iam.gserviceaccount.com>
This reverts commit e5a06ce678.
Reason for revert: Need to make change in google3 first
Original change's description:
> Move GrGpuResource GrSurface and GrTexture into src.
>
> Must land https://chromium-review.googlesource.com/c/chromium/src/+/2087980
> before this can land.
>
> Bug: skia:7966
> Change-Id: I60bbb1765bfbb2c96b2bc0c9826b6b9d57eb2a03
> Reviewed-on: https://skia-review.googlesource.com/c/skia/+/275077
> Commit-Queue: Greg Daniel <egdaniel@google.com>
> Reviewed-by: Robert Phillips <robertphillips@google.com>
TBR=egdaniel@google.com,bsalomon@google.com,robertphillips@google.com
Change-Id: Id39e0a351e49a87209de88a6ad9fadb0219db72c
No-Presubmit: true
No-Tree-Checks: true
No-Try: true
Bug: skia:7966
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/275216
Reviewed-by: Greg Daniel <egdaniel@google.com>
Commit-Queue: Greg Daniel <egdaniel@google.com>
Change-Id: I3903e266e2e99ceab647fbc45c764e3526e600af
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/275198
Commit-Queue: Mike Klein <mtklein@google.com>
Commit-Queue: Herb Derby <herb@google.com>
Reviewed-by: Herb Derby <herb@google.com>
Count pixels, show percents, and promote to int64_t
for safety... pixels certainly needs it.
Change-Id: Ia8db4fec6c357557e6c805b6c05191f71e331843
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/275148
Commit-Queue: Herb Derby <herb@google.com>
Reviewed-by: Herb Derby <herb@google.com>
They'll never see fma_f32 ops.
Change-Id: I39371606c673fb76bdcbbe08c1b25308675f8f2c
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/275151
Commit-Queue: Mike Klein <mtklein@google.com>
Reviewed-by: Mike Klein <mtklein@google.com>
This version has fewer uninitentional sync points.
Turning on stats, looks like roughly 93% of calls
go through the JIT, the rest the interpreter. When
synchronous 100% go through the JIT, of course.
No diffs now that we've made the interpreter and JIT
guarantee consistent math.
Change-Id: Id58c860809970f78773b2ae7f40b3ea091b38440
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/275011
Commit-Queue: Mike Klein <mtklein@google.com>
Commit-Queue: Herb Derby <herb@google.com>
Reviewed-by: Herb Derby <herb@google.com>
This helps guarantee same-machine consistency
between the interpreter and the JIT.
2-3% perf loss on exemplar benches, which we think we can
recover with the following TODO:
TODO(herb): add Op::{fnma,fms,fnms}_f32 for the various ±xy ± z
combinations we can generate in the backends, peephole, etc.
just as we do Op::fma_f32 so we can recover the (tiny) performance
loss here we take from losing those discovered fused instructions.
Change-Id: Id30e1b9178a55e8072a30d2a34e7ed9188543d60
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/275139
Commit-Queue: Herb Derby <herb@google.com>
Reviewed-by: Herb Derby <herb@google.com>
Peephole add(F32,F32) for an argument that is a mul().
As a flourish, only generate Op::fma_f32 on machines we know support
real fused mul-adds. This removes the ambiguity of whether Op::mad_f32
is an FMA or not; the new Op::fma_f32 is always an FMA, and otherwise
you'll just see ordinary mul-add. No more Op::mad_f32.
Change-Id: I38016a2430774583116d8d6a8ada677012c1a8fc
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/275138
Reviewed-by: Herb Derby <herb@google.com>
Commit-Queue: Herb Derby <herb@google.com>
We really only need to_unorm(),
and that's fine with trunc(mad(x, scale, 0.5)).
Change-Id: I1561c678501963a9ae53c22994fc906159fc7199
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/275075
Commit-Queue: Mike Klein <mtklein@google.com>
Reviewed-by: Herb Derby <herb@google.com>
Reviewed-by: Mike Klein <mtklein@google.com>
I used the wrong data type for encoding the alt text and language of
a node in the PDF structure tree. "Alt" and "Lang" should both be
strings (arbitrary text), not names (which are supposed to be used for
keywords that are defined as part of the text).
Bug: chromium:607777
Change-Id: I1000c177d66a1e4f68afc713fc950369ecf3b408
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/275010
Reviewed-by: Ben Wagner <bungeman@google.com>
Commit-Queue: Dominic Mazzoni <dmazzoni@chromium.org>
This fixes a bug where a finish proc from the checkCommandBuffer call could
allow the client to unref an SkImage/SkSurface which may internally cause
us to try to add a command to a commandBuffer (e.g. to switch queues).
Bug: chromium:1049864
Change-Id: I975aa0b17acfdf0df30c48af685f9e6753f095db
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/275042
Reviewed-by: Jim Van Verth <jvanverth@google.com>
Commit-Queue: Greg Daniel <egdaniel@google.com>
The move constructor and operator= were unfortunate sync points when
trying to do async compilation because fJITEntry was about to move.
We can fix this by making the jit entry target not move even when
Program does, by
- banning Program moves, replacing uses of Program that need that
with std::unique_ptr<Program>,
- or equivalently, doing that internally with a pimpl pattern.
There was enough moving of Programs that I think the pimpl is the
better choice, at least for now.
Change-Id: I014775c473662d9901ba387c48b25fa75ad7a078
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/275006
Commit-Queue: Mike Klein <mtklein@google.com>
Reviewed-by: Herb Derby <herb@google.com>
Fix for MSAN in tilemode_decal?
Change-Id: I58df5a5e1fdd5a16737d59ad0863df9ca9b15920
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/275040
Auto-Submit: Brian Salomon <bsalomon@google.com>
Commit-Queue: Brian Salomon <bsalomon@google.com>
Commit-Queue: Michael Ludwig <michaelludwig@google.com>
Reviewed-by: Michael Ludwig <michaelludwig@google.com>
Bug: skia:9935
Change-Id: I6d9c14ac674dee89a5a8ba726d99b4423c40eb9b
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/274861
Reviewed-by: Eric Boren <borenet@google.com>
Reviewed-by: Greg Daniel <egdaniel@google.com>
Commit-Queue: Jim Van Verth <jvanverth@google.com>
Documentation specifies that only the alpha, color filter, image filter
and blend mode are used when restoring the saved layer. The coverage that
would be sent to the mask filter is ill-defined, whereas the explicit
boundary of the separate clip image gets around this.
This also removes the issue of the order in which the mask filter and
image filter would be applied.
Some GMs exercised the save layer with mask filter, but those have
been updated to either expect the MF to be ignored, or modified to
no longer use saveLayer+MF. While Android and Flutter expose a saveLayer
with general Paint API, their documentation matches SkCanvas so hopefully
no one is actually relying on the mask filter in those cases. It does
not appear as if Chrome ever tries to use a saveLayer+MF.
Bug: skia:9561
Change-Id: I67a350fbf565683ea60ea88f83a91270e7dcd187
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/274508
Reviewed-by: Mike Reed <reed@google.com>
Commit-Queue: Michael Ludwig <michaelludwig@google.com>
In support of this add support for border color to GrTextureEffect.
Currently non-zero border colors are always emulated in shader code.
Change-Id: I007e264411d713f8afaf8f160b6cea6d0f35c753
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/274282
Reviewed-by: Michael Ludwig <michaelludwig@google.com>
Commit-Queue: Brian Salomon <bsalomon@google.com>
b84969ad69..9e9493f29d
git log b84969ad69c3..9e9493f29dbf --date=short --first-parent --format='%ad %ae %s'
2020-03-04 ynovikov@chromium.org Skip crashing dEQP-GLES3.functional.samplers.*_tex_3d.* on SwANGLE
2020-03-03 cnorthrop@google.com Capture/Replay: Add Query Object support to mid-execution capture
2020-03-03 cnorthrop@google.com Capture/Replay: Update getTexImage for depth/stencil and 3D
2020-03-03 ancheng.qiao@arm.com Vulkan: Add depth to mipmap generation
2020-03-03 xiaoxuan.liu@arm.com Vulkan: Add support for VK_EXT_index_type_uint8
2020-03-03 angle-autoroll@skia-public.iam.gserviceaccount.com Roll third_party/SwiftShader 416d37452275..176f3a3dbbc9 (2 commits)
2020-03-03 angle-autoroll@skia-public.iam.gserviceaccount.com Roll third_party/vulkan-validation-layers/src cb689b0b1d30..f32bf1c930c8 (8 commits)
Created with:
gclient setdep -r third_party/externals/angle2@9e9493f29dbf
If this roll has caused a breakage, revert this CL and stop the roller
using the controls here:
https://autoroll.skia.org/r/angle-skia-autoroll
Please CC herb@google.com on the revert to ensure that a human
is aware of the problem.
To report a problem with the AutoRoller itself, please file a bug:
https://bugs.chromium.org/p/skia/issues/entry?template=Autoroller+Bug
Documentation for the AutoRoller is here:
https://skia.googlesource.com/buildbot/+/master/autoroll/README.md
Cq-Include-Trybots: skia/skia.primary:Build-Debian9-Clang-x86_64-Release-ANGLE;skia/skia.primary:Test-Win10-Clang-AlphaR2-GPU-RadeonR9M470X-x86_64-Debug-All-ANGLE;skia/skia.primary:Test-Win10-Clang-Golo-GPU-QuadroP400-x86_64-Debug-All-ANGLE;skia/skia.primary:Test-Win10-Clang-NUC5i7RYH-GPU-IntelIris6100-x86_64-Debug-All-ANGLE;skia/skia.primary:Test-Win10-Clang-NUC6i5SYK-GPU-IntelIris540-x86_64-Debug-All-ANGLE;skia/skia.primary:Test-Win10-Clang-NUC8i5BEK-GPU-IntelIris655-x86_64-Debug-All-ANGLE;skia/skia.primary:Test-Win10-Clang-NUCD34010WYKH-GPU-IntelHD4400-x86_64-Debug-All-ANGLE
Bug: None
Tbr: herb@google.com
Change-Id: I349f021a8ad84dde2afb8fc65de96f41d6100364
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/274976
Reviewed-by: skia-autoroll <skia-autoroll@skia-public.iam.gserviceaccount.com>
Commit-Queue: skia-autoroll <skia-autoroll@skia-public.iam.gserviceaccount.com>
This reverts commit a624a534ed.
Reason for revert: Bad perf
Original change's description:
> Use spin lock in SkIDChangeListener
>
> This lock should almost never be contested. This simplifies things and
> also avoid mutex locks when one thread has multiple refs on a path,
> image, ...
>
> Change-Id: I909b490363cb9e81b38ed9edd04272133fb2692b
> Reviewed-on: https://skia-review.googlesource.com/c/skia/+/274676
> Reviewed-by: Brian Osman <brianosman@google.com>
> Commit-Queue: Brian Salomon <bsalomon@google.com>
TBR=bsalomon@google.com,brianosman@google.com
Change-Id: I45ec3241906429e4d0783b68ebe6398079b73d36
No-Presubmit: true
No-Tree-Checks: true
No-Try: true
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/274956
Reviewed-by: Brian Salomon <bsalomon@google.com>
Commit-Queue: Brian Salomon <bsalomon@google.com>
It already makes copies for other reason and is the only generator
relying on the caller.
Change-Id: I5dfdc3bd39040f817c0f953eecf81e8fbdc649a4
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/274862
Commit-Queue: Brian Salomon <bsalomon@google.com>
Reviewed-by: Brian Osman <brianosman@google.com>
With the addition programInfos being stored on DDLs we need to ensure that the programInfos are deleted before the GrCCPerOpsTaskPaths.
W/o this change we can hit an assert in ~GrCCClipPath.
Change-Id: I02a2b195cae1dfd84a3e9cd2cf6c9b5194c430a8
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/274860
Reviewed-by: Brian Salomon <bsalomon@google.com>
Commit-Queue: Robert Phillips <robertphillips@google.com>
Change-Id: I7ae6100a36c1764f40b97f089817b77263544e8a
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/274857
Auto-Submit: Stephen White <senorblanco@chromium.org>
Commit-Queue: Brian Salomon <bsalomon@google.com>
Reviewed-by: Brian Salomon <bsalomon@google.com>
As far as I can tell the TLS contexts were an unjustified paranoia.
90%+ of runtime in setupLLVM() is spent under getFunctionAddress()
JITting the function. This change does not affect overhead AFAICT.
Change-Id: I2abe3832ed7b42c0ac67a7605cf7d0bf63e7bd55
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/274783
Commit-Queue: Mike Klein <mtklein@google.com>
Reviewed-by: Herb Derby <herb@google.com>
