Also add GrDawnCaps to friends of GrShaderCaps, as the other backends do.
Change-Id: Iae60cae7168ad9194f7808c2f51f41de10e53071
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/230747
Commit-Queue: Stephen White <senorblanco@chromium.org>
Reviewed-by: Greg Daniel <egdaniel@google.com>
Change-Id: I49a233a28d16622b286ffc4150446b33f899a891
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/230557
Reviewed-by: Greg Daniel <egdaniel@google.com>
Commit-Queue: Stephen White <senorblanco@chromium.org>
Change-Id: Ic0fd9f3bb807a8377a63a07321e26c4ee5cdeaa7
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/231176
Reviewed-by: Brian Salomon <bsalomon@google.com>
Commit-Queue: Stephen White <senorblanco@chromium.org>
The goal here is to help reduce a lot of the refs and unrefs that happen
during each of our draws.
Change-Id: I81b45a2259acf24b73d1a01f8574a48c0adae9fa
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/230741
Reviewed-by: Jim Van Verth <jvanverth@google.com>
Commit-Queue: Greg Daniel <egdaniel@google.com>
Also add a unit test that crashes before this change.
Change-Id: I94e441a57a9c28e7c12bc2b214a65b41446ffab8
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/230754
Reviewed-by: Ben Wagner <bungeman@google.com>
Commit-Queue: Hal Canary <halcanary@google.com>
Add weird color types that handle the swizzling.
Change-Id: Ie37a00eb877fe5e519f7498bf749e02a2f1dc204
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/230135
Reviewed-by: Greg Daniel <egdaniel@google.com>
Commit-Queue: Brian Salomon <bsalomon@google.com>
This omission appears to have been causing YUV problems in Chrome.
Change-Id: I7f13b18ba5c367dcdc242eff3532cacb2e06d1da
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/231100
Reviewed-by: Brian Salomon <bsalomon@google.com>
Commit-Queue: Robert Phillips <robertphillips@google.com>
This is in preparation of exposing new factory functions DropShadow()
and DropShadowOnly(), which do not take the enum at all in their params.
Bug: skia:9280
Change-Id: If6e16f46e94b54ab156d2fd02c59fd6171a30f2e
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/230742
Commit-Queue: Michael Ludwig <michaelludwig@google.com>
Reviewed-by: Robert Phillips <robertphillips@google.com>
Change-Id: Ibd541b4f4d8a9ab2b86743c86a6cfbc50a56518f
TBR: benjaminwagner@google.com
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/230891
Reviewed-by: Ben Wagner aka dogben <benjaminwagner@google.com>
Reviewed-by: Michael Ludwig <michaelludwig@google.com>
Commit-Queue: Michael Ludwig <michaelludwig@google.com>
Fixes problems when adding user-defined helper functions to the SkSL.
Change-Id: I00c3031f67e70039c09ac375774778b698d8a3c8
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/230889
Commit-Queue: Brian Osman <brianosman@google.com>
Commit-Queue: Mike Klein <mtklein@google.com>
Auto-Submit: Brian Osman <brianosman@google.com>
Reviewed-by: Mike Klein <mtklein@google.com>
Just because a program doesn't read from or
write to memory doesn't mean it's pointless.
Oh wait, yes it does. It shouldn't crash though.
Change-Id: I6a9c26c065831f9598afccce6e0a34a178cbd925
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/230839
Reviewed-by: Mike Klein <mtklein@google.com>
Commit-Queue: Mike Klein <mtklein@google.com>
This reverts commit 1dfb4d25d1.
Reason for revert: This appears to have poorer performance on integrated GPUs.
Original change's description:
> Use shared mode for Metal texture upload buffers.
>
> On Mac, originally the temporary upload buffer was stored in managed
> memory, which will generate two copy actions, one from the CPU side to
> the GPU side of the managed buffer, then another from the GPU side of
> the buffer to the GPU texure. Shared mode buffers are in PCI-accessible
> system memory, so by storing the temporary texture data there, we should
> be copying from the CPU buffer directly to the GPU texture. This gives
> us another 10-15% perf improvement over the old upload code.
>
> Change-Id: Ia675e7130e0ae7d0c5f16c8170eeef60a69bfbbd
> Reviewed-on: https://skia-review.googlesource.com/c/skia/+/230299
> Reviewed-by: Brian Osman <brianosman@google.com>
> Commit-Queue: Jim Van Verth <jvanverth@google.com>
TBR=egdaniel@google.com,jvanverth@google.com,brianosman@google.com
# Not skipping CQ checks because original CL landed > 1 day ago.
Change-Id: Ib89b014586b7bbd899fed5d8e5807e083368a825
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/230882
Reviewed-by: Jim Van Verth <jvanverth@google.com>
Commit-Queue: Jim Van Verth <jvanverth@google.com>
Fixes issue where we pass an invalid format, GL_SRGB_ALPHA, when
allocating render buffers.
Change-Id: Ie4ea9f4a4edb2d525b1c32d6a4f8a94c4b55cdb5
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/230751
Reviewed-by: Greg Daniel <egdaniel@google.com>
Commit-Queue: Brian Salomon <bsalomon@google.com>
Change-Id: I00276862596343456519bdb506e74c355d1834a8
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/230877
Commit-Queue: Greg Daniel <egdaniel@google.com>
Reviewed-by: Brian Osman <brianosman@google.com>
Fills in a little TODO.
Also add assembler unit tests for similar float->int and int->float.
Tested by SkVM_mad, SkVM_madder.
Change-Id: I5334029927fdecb0ff7f5a3b081cf2ce7b23995c
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/230838
Reviewed-by: Mike Klein <mtklein@google.com>
Commit-Queue: Mike Klein <mtklein@google.com>
Bug: skia:9280
Change-Id: Ice3624467f5e643f9d3d81da72d77911eeac963e
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/230198
Commit-Queue: Michael Ludwig <michaelludwig@google.com>
Reviewed-by: Mike Reed <reed@google.com>
Reviewed-by: Robert Phillips <robertphillips@google.com>
This makes calling GrCaps::getBackendFormatFromColorType equivalent to calling GrContext::defaultBackendFormat
Change-Id: Ic402245fa0502920c47c2a29aadc4e559abdc09b
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/230416
Commit-Queue: Robert Phillips <robertphillips@google.com>
Reviewed-by: Greg Daniel <egdaniel@google.com>
Added virtual transferFrom to GrOpList and implemented in both
subclasses.
Bug: skia:9269
Change-Id: Ic930e46b172879e592897512c81a0cfe79f20d19
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/230129
Reviewed-by: Brian Osman <brianosman@google.com>
Commit-Queue: Brian Salomon <bsalomon@google.com>
Removes an unnecessary use of GrPixelConfig.
Bug: skia:6718
Change-Id: Ib020c42eec58fc53aaf5034fb37bf9d6ec459f95
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/230748
Reviewed-by: Greg Daniel <egdaniel@google.com>
Commit-Queue: Brian Salomon <bsalomon@google.com>
Also fix some whitespace issues.
Change-Id: Ifa2bb7c421604c37861ae0726c6b2aabfefdec4f
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/230558
Reviewed-by: Greg Daniel <egdaniel@google.com>
Commit-Queue: Stephen White <senorblanco@chromium.org>
Change-Id: I58b52d3e1d05d0834be30e00d991636e227cbf0b
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/230836
Reviewed-by: Mike Klein <mtklein@google.com>
Commit-Queue: Mike Klein <mtklein@google.com>
Change-Id: I76824eca6e72505004b4dfb03fc35c2849176ecb
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/230459
Commit-Queue: Stephen White <senorblanco@chromium.org>
Reviewed-by: Greg Daniel <egdaniel@google.com>
Bug: skia:9280
Change-Id: I638c2b3d34be2037648cf27bde933345f8abbd77
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/230126
Commit-Queue: Michael Ludwig <michaelludwig@google.com>
Reviewed-by: Robert Phillips <robertphillips@google.com>
- drop unused ctor
- all layers are kPremul_SkAlphaType -> drop preserveLCDText logic
Change-Id: I24b9d3273ac65d9775714272a75222f36e44a453
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/227643
Reviewed-by: Mike Reed <reed@google.com>
Commit-Queue: Florin Malita <fmalita@chromium.org>
This is part of trying to remove uses of GrPixelConfig from the proxy provider
Change-Id: I12d085cfbff86d0e44829ce3ee36744b937b804e
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/230576
Reviewed-by: Greg Daniel <egdaniel@google.com>
Commit-Queue: Robert Phillips <robertphillips@google.com>
We can't remove the loopers themselves, as they are still used
by android and chrome (they just don't ever pass them to skia).
Eventually each of those clients will resolve this, but for now
we just keep the classes (and tests) in skia.
Bug: skia:4783
Change-Id: I5f507e6bb82280f2bc7c0b21eebe59c287aa9265
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/230579
Reviewed-by: Florin Malita <fmalita@chromium.org>
Commit-Queue: Mike Reed <reed@google.com>
This first implementation calls SetSubData() to upload, and simply
uses malloc'ed buffers for onMap()/onUnmap().
Change-Id: I433b4b094a77fbfe1a21447127795b46fdfb9803
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/230458
Commit-Queue: Stephen White <senorblanco@chromium.org>
Reviewed-by: Greg Daniel <egdaniel@google.com>
Add missing comparison and selection ops, bit casts, 16-bit memory
operations, gathers, uniform loads, and fill in math holes where
reasonable. Update some names to be a bit more regular.
I think all instructions are implemented in the interpreter,
and many tested. More testing and JITs to follow.
Change-Id: I8cf377e8b72a86ac950e020892ce82b39e9d7277
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/229893
Reviewed-by: Mike Klein <mtklein@google.com>
Commit-Queue: Mike Klein <mtklein@google.com>
Change-Id: Ib37cc549bfd52a5fb3f1f8ef76d0117ed2a1f43d
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/230419
Reviewed-by: Jim Van Verth <jvanverth@google.com>
Commit-Queue: Greg Daniel <egdaniel@google.com>
Bug: skia:9280
Change-Id: I6d40b83093d8a8e9e84bbe44d917d02dfde85877
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/230124
Commit-Queue: Michael Ludwig <michaelludwig@google.com>
Reviewed-by: Robert Phillips <robertphillips@google.com>
Bug: skia:
Change-Id: I2ae0caf08f8434302cae8151ae1ea0fda8d56928
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/230397
Reviewed-by: Brian Osman <brianosman@google.com>
Commit-Queue: Ethan Nicholas <ethannicholas@google.com>
On Mac, originally the temporary upload buffer was stored in managed
memory, which will generate two copy actions, one from the CPU side to
the GPU side of the managed buffer, then another from the GPU side of
the buffer to the GPU texure. Shared mode buffers are in PCI-accessible
system memory, so by storing the temporary texture data there, we should
be copying from the CPU buffer directly to the GPU texture. This gives
us another 10-15% perf improvement over the old upload code.
Change-Id: Ia675e7130e0ae7d0c5f16c8170eeef60a69bfbbd
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/230299
Reviewed-by: Brian Osman <brianosman@google.com>
Commit-Queue: Jim Van Verth <jvanverth@google.com>
Use that to implement many intrinsics in SkSL for the interpreter.
Change-Id: I3762867781cb1a053429fd37b12ae3cf7739cb3c
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/230134
Reviewed-by: Ethan Nicholas <ethannicholas@google.com>
Reviewed-by: Mike Klein <mtklein@google.com>
Commit-Queue: Brian Osman <brianosman@google.com>
Change-Id: Ib442d085bc51b10a11ba575bdcee532931c5a32b
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/230257
Commit-Queue: Stephen White <senorblanco@chromium.org>
Reviewed-by: Robert Phillips <robertphillips@google.com>
There's a fair bit of overhead to setting up a GrMtlBuffer (adding it
to the resource cache, etc.) that isn't necessary in this case when
using a temporary buffer. Overall it gives about 10% improvement in
uploadToTexture().
Change-Id: I99449c9001098c90743cba8d614007db307156d3
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/230131
Commit-Queue: Jim Van Verth <jvanverth@google.com>
Reviewed-by: Greg Daniel <egdaniel@google.com>
Bug: skia:6718
Change-Id: Ifd78a1b227dc468ce882cf3caf78b4fae80cc006
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/229386
Commit-Queue: Robert Phillips <robertphillips@google.com>
Reviewed-by: Greg Daniel <egdaniel@google.com>
Reviewed-by: Jim Van Verth <jvanverth@google.com>
Dawn doesn't support old-style combined texturesampler uniforms, so
they must be expressed as separate samplers and texture uniforms.
At the SkSL Type level, I've added a Texture2D type, and expressed
the Sampler2D (combined) type in terms of it. This ensures that we
emit only a single OpTypeImage for it in the SPIRV.
Eventually, all of the Texture types (1D, 3D, Rect) could be defined
and SamplerX could simply contain a reference to TextureX. I wanted to
float this idea with a single example for now (and since it's all that
the Dawn backend needs).
This also required adding a new "makeSampler2D" function to combine
them, which maps to OpSampledImage at the SPIR-V level.
Change-Id: Iaf33a6e7d339da415be6ea9a017340cb0ef3c1eb
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/229417
Commit-Queue: Stephen White <senorblanco@chromium.org>
Reviewed-by: Ethan Nicholas <ethannicholas@google.com>
It's possible to hit this with the comma operator. Not sure if there are
any others.
Change-Id: I10fb68d7a20ce54fccc1597da8e656fa7d61cd97
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/230197
Reviewed-by: Brian Osman <brianosman@google.com>
Commit-Queue: Brian Osman <brianosman@google.com>
The recent change to output child processors as functions served as a
stress test for Metal functions and uncovered some issues with
requirements tracking. This should get everything back on track.
Bug: skia:
Change-Id: If683c95e6668f1afd7be24195071a26cdb84d660
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/230130
Reviewed-by: Brian Osman <brianosman@google.com>
Reviewed-by: Jim Van Verth <jvanverth@google.com>
Commit-Queue: Ethan Nicholas <ethannicholas@google.com>
Bug: skia:9280
Change-Id: I1a1df193ce212300b7a84d2247fa9c63601c9446
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/230119
Commit-Queue: Michael Ludwig <michaelludwig@google.com>
Reviewed-by: Robert Phillips <robertphillips@google.com>
Change-Id: I0773bb6d91bc14cdb528892a8b28fd9af657153b
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/230132
Commit-Queue: Greg Daniel <egdaniel@google.com>
Commit-Queue: Brian Salomon <bsalomon@google.com>
Auto-Submit: Greg Daniel <egdaniel@google.com>
Reviewed-by: Brian Salomon <bsalomon@google.com>
Very likely, reimplementing this routine using doubles would also
improve results, given the many divides, subtracts, and trig calls.
Bug: skia:9272
Change-Id: I78a3dd9a11a8de1c0ef7a4406a227a0b7d9aecf7
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/230123
Reviewed-by: Florin Malita <fmalita@chromium.org>
Commit-Queue: Mike Reed <reed@google.com>
Bug: skia:9278
Change-Id: I3196ccabb5c8f685f050d881548592a7c16fbec2
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/230117
Reviewed-by: Eric Boren <borenet@google.com>
Commit-Queue: Chris Dalton <csmartdalton@google.com>
This reverts commit e2c5e8c7ee.
Reason for revert: relanding with fix
Original change's description:
> Revert "Move ExternalFormat and Type to FormatInfo."
>
> This reverts commit 80140518ef.
>
> Reason for revert: probably breaking angle
>
> Original change's description:
> > Move ExternalFormat and Type to FormatInfo.
> >
> > This also deletes the ConfigInfoTable in GrGLCaps as there is no more use
> > of it.
> >
> > Additionally with the rework of storing External Io info on the format table
> > I rewrote the implimination of supportedReadPixels and supportedWritePixels
> > for GL to loop over the supported types looking for a match instead of
> > simply defaulting to a base value.
> >
> > Finally transferFromOffsetAlignment has been rolled into the SupportedRead
> > instead of being its own query.
> >
> > Bug: skia:6718
> > Change-Id: I39f77adf6c0b5b38245e55e8a7e18c0b428862d0
> > Reviewed-on: https://skia-review.googlesource.com/c/skia/+/229381
> > Commit-Queue: Greg Daniel <egdaniel@google.com>
> > Reviewed-by: Brian Salomon <bsalomon@google.com>
>
> TBR=egdaniel@google.com,bsalomon@google.com,robertphillips@google.com
>
> Change-Id: Ifef2e7308fdb4d91d649f08488b798815e0aa5fa
> No-Presubmit: true
> No-Tree-Checks: true
> No-Try: true
> Bug: skia:6718
> Reviewed-on: https://skia-review.googlesource.com/c/skia/+/229896
> Reviewed-by: Greg Daniel <egdaniel@google.com>
> Commit-Queue: Greg Daniel <egdaniel@google.com>
TBR=egdaniel@google.com,bsalomon@google.com,robertphillips@google.com
Change-Id: Ida09be706b461cf89467fc0082744177e71e8985
Bug: skia:6718
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/229918
Reviewed-by: Greg Daniel <egdaniel@google.com>
Reviewed-by: Brian Salomon <bsalomon@google.com>
Commit-Queue: Greg Daniel <egdaniel@google.com>
This reverts commit 80140518ef.
Reason for revert: probably breaking angle
Original change's description:
> Move ExternalFormat and Type to FormatInfo.
>
> This also deletes the ConfigInfoTable in GrGLCaps as there is no more use
> of it.
>
> Additionally with the rework of storing External Io info on the format table
> I rewrote the implimination of supportedReadPixels and supportedWritePixels
> for GL to loop over the supported types looking for a match instead of
> simply defaulting to a base value.
>
> Finally transferFromOffsetAlignment has been rolled into the SupportedRead
> instead of being its own query.
>
> Bug: skia:6718
> Change-Id: I39f77adf6c0b5b38245e55e8a7e18c0b428862d0
> Reviewed-on: https://skia-review.googlesource.com/c/skia/+/229381
> Commit-Queue: Greg Daniel <egdaniel@google.com>
> Reviewed-by: Brian Salomon <bsalomon@google.com>
TBR=egdaniel@google.com,bsalomon@google.com,robertphillips@google.com
Change-Id: Ifef2e7308fdb4d91d649f08488b798815e0aa5fa
No-Presubmit: true
No-Tree-Checks: true
No-Try: true
Bug: skia:6718
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/229896
Reviewed-by: Greg Daniel <egdaniel@google.com>
Commit-Queue: Greg Daniel <egdaniel@google.com>
This also deletes the ConfigInfoTable in GrGLCaps as there is no more use
of it.
Additionally with the rework of storing External Io info on the format table
I rewrote the implimination of supportedReadPixels and supportedWritePixels
for GL to loop over the supported types looking for a match instead of
simply defaulting to a base value.
Finally transferFromOffsetAlignment has been rolled into the SupportedRead
instead of being its own query.
Bug: skia:6718
Change-Id: I39f77adf6c0b5b38245e55e8a7e18c0b428862d0
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/229381
Commit-Queue: Greg Daniel <egdaniel@google.com>
Reviewed-by: Brian Salomon <bsalomon@google.com>
Change-Id: Ibb637415a53510027838c23cc2316665ffbd6c82
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/209150
Reviewed-by: Mike Klein <mtklein@google.com>
Commit-Queue: Mike Klein <mtklein@google.com>
Auto-Submit: Mike Klein <mtklein@google.com>
PS2 adds a rewrite for Skia #include <...> to #include "...", letting
them be otherwise rewritten and sorted too. (We do need one exception
for the Vulkan headers, which will otherwise be rewritten to always
point to our own.) I don't think it's particularly important to
favor "" or <>, but picking one keeps things consistent.
PS3 adds a missing SkMutex.h include.
PS4 fixes a terrible readability problem.
Change-Id: Id9fe752727ef30e802b1daf755ee2ed15e267577
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/229742
Commit-Queue: Mike Klein <mtklein@google.com>
Commit-Queue: Brian Osman <brianosman@google.com>
Auto-Submit: Mike Klein <mtklein@google.com>
Reviewed-by: Brian Osman <brianosman@google.com>
This member variable is no longer used
TBR=bsalomon@google.com
Bug: skia:6718
Change-Id: I84680c8c3bc36eefa603f5be7f3c15b496a79948
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/229478
Reviewed-by: Robert Phillips <robertphillips@google.com>
Reviewed-by: Greg Daniel <egdaniel@google.com>
Commit-Queue: Robert Phillips <robertphillips@google.com>
Bug: skia:
Change-Id: Icbf8f542637a874b3e2d3513d932b39728fa5e77
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/229385
Reviewed-by: Brian Osman <brianosman@google.com>
Reviewed-by: Brian Salomon <bsalomon@google.com>
Commit-Queue: Ethan Nicholas <ethannicholas@google.com>
Change-Id: I2c76d8cbcfc3f36448127de5a3e1a22f76eda863
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/229489
Reviewed-by: Mike Klein <mtklein@google.com>
Commit-Queue: Brian Osman <brianosman@google.com>
Bug: chromium:985129
Change-Id: I7dfea73f7eb48aeba11c110fc8b9fa253770afc1
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/229493
Commit-Queue: Jim Van Verth <jvanverth@google.com>
Commit-Queue: Brian Salomon <bsalomon@google.com>
Auto-Submit: Jim Van Verth <jvanverth@google.com>
Reviewed-by: Brian Salomon <bsalomon@google.com>
A 180 degree rotation matrix is functionally like a scale matrix, but
the elements are negative. To compensate for this, we use abs().
Bug: b/137547660
Change-Id: Ib0e7449872523af024e7de9005b9bb70743e04b5
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/229394
Commit-Queue: Jim Van Verth <jvanverth@google.com>
Reviewed-by: Robert Phillips <robertphillips@google.com>
Instead we leave color writes enabled and use a blend state that
preserves the dst color. This allows us to re-enable msaa ccpr on
PowerVR.
Bug: skia:
Change-Id: I1e902d695ad483ffb13dff6a7920749e307b49c7
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/229387
Commit-Queue: Chris Dalton <csmartdalton@google.com>
Reviewed-by: Brian Salomon <bsalomon@google.com>
Note: I'll follow this up with a separate CL that removes the pixel config
Bug: skia:6718
Change-Id: If069afa95bd51d5d6b24089fd3a8526e4d982820
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/228257
Reviewed-by: Brian Salomon <bsalomon@google.com>
Commit-Queue: Robert Phillips <robertphillips@google.com>
As written it was hoisting one register per tbl instruction use, which
was both correct and fast, but using more registers than necessary. We
only need one table mask for each unique Op::bytes control value.
Caught this looking at the profile for srcover_i32, which uses each of
two distinct bytes control values twice, once for the src pixel and once
for dst.
Change-Id: I42f8611b23c0258369e8aaadd75fc484d1e9cb35
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/229397
Reviewed-by: Mike Klein <mtklein@google.com>
Commit-Queue: Mike Klein <mtklein@google.com>
fix toy square program
add a new toy plus_one program
also run stock srcover programs
time at N=15,255,4095
e.g.
plus_one 69 Mpx/s 435 Mpx/s 615 Mpx/s
square 73 Mpx/s 393 Mpx/s 527 Mpx/s
srcover_f32 19 Mpx/s 38 Mpx/s 39 Mpx/s
srcover_i32 36 Mpx/s 89 Mpx/s 103 Mpx/s
srcover_i32_naive 34 Mpx/s 85 Mpx/s 91 Mpx/s
srcover_i32_SWAR 49 Mpx/s 169 Mpx/s 191 Mpx/s
Change-Id: Ie041603bf1b458472dd477c64a91c61162a362e3
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/229396
Commit-Queue: Mike Klein <mtklein@google.com>
Reviewed-by: Mike Klein <mtklein@google.com>
This was effectively happening before, but only after we (always) failed
to find an entry without the origin. If we ever have a system that can
determine ahead of time that we won't depend on surface origin, we can
update this logic. For now, just simplify the code and remove the
useless cache lookups.
Change-Id: I2b6178a8a83a48f33fdc066e185a2a4c13aa8ddc
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/229284
Reviewed-by: Brian Salomon <bsalomon@google.com>
Commit-Queue: Brian Osman <brianosman@google.com>
This CL is intended to further wean Ganesh off of using the GrBackendTexture's pixel config
TBR=bsalomon@google.com
Bug: skia:6718
Change-Id: Iedaa7811f9c4aac552f219c702627bc476325317
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/228338
Reviewed-by: Robert Phillips <robertphillips@google.com>
Reviewed-by: Greg Daniel <egdaniel@google.com>
Commit-Queue: Robert Phillips <robertphillips@google.com>
Bug: skia:9269
Change-Id: I7179c298f7b5ae580889d8fa7e3821fb2408aaad
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/229077
Reviewed-by: Brian Osman <brianosman@google.com>
Commit-Queue: Brian Salomon <bsalomon@google.com>
Calling add32 can re-allocate the key storage, so it isn't safe to call
add32n and then keep writing into the returned pointer if an extra
sampler key is encountered.
Also fix the GP sampler key code to actually set the last bit, not the
16th bit.
Change-Id: I9c83435a164ab0391e2e6a9e1a9bdf6839490a15
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/229278
Reviewed-by: Greg Daniel <egdaniel@google.com>
Commit-Queue: Brian Osman <brianosman@google.com>
This leaves just width, height, and GrPixelConfig. Once we remove the
latter we can replace GrSurfaceDesc with SkISize.
Also remove unused GrRenderTarget::overrideResolveRect
Also remove GrSurfaceProxy::Renderable and use GrRenderable instead.
Change-Id: I652fe6169a22ca33d199b144ec6385286ac07b5a
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/228570
Reviewed-by: Greg Daniel <egdaniel@google.com>
Commit-Queue: Brian Salomon <bsalomon@google.com>
Adding cache
Caching shaped results
Base+Index for referencing arrays
The very first and naive version of cache
Cache measurement, lines and picture
Added text blob cache for lines
Removed Run* from Cluster
Removed const char* from Cluster and Run
Few minor changes
Change-Id: I444a1defa950aed5999cfa1c3545fd83ccb54ce9
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/227840
Commit-Queue: Julia Lavrova <jlavrova@google.com>
Reviewed-by: Julia Lavrova <jlavrova@google.com>
Reviewed-by: Mike Reed <reed@google.com>
This adds a warmup phase to let each instruction do any setup
it needs, adding lookup entries for splat and bytes, and
on aarch64, hoisting the mask to a register when we can.
Oddly, this measures as a ~3x slowdown on the phone I'm testing, an
international Galaxy S9 with a Samsung Mongoose 3 processor. I've got
to imagine this somehow makes the processor think there's a carried loop
dependency when there is not? Anyway, we already know that that's a
pretty crazy CPU (reports FP16 compute but cannot), and this does deliver
a speedup on the Pixel 2's Kryo 280 / Cortex A73, so I think maybe I'll
just swap back to testing with the Pixel 2 and forget about that S9.
Here's a before/after codelisting with a hoisted tbl mask. In the
before case it's loaded in the loop with `ldr q3, #152`, and becomes
`ldr q0, #168` outside the loop. llvm-mca says this should cut one
cycle per loop, and with optimal out of order execution the loop cost
would drop from ~8.7 cycles to ~8.3. In practice, it looks like about a
15% speedup.
before:
ldr q0, #188
ldr q1, #200
cmp x0, #4 // =4
b.lt #76
ldr q2, [x1]
ldr q3, #152
tbl v3.16b, { v2.16b }, v3.16b
sub v3.8h, v0.8h, v3.8h
ldr q4, [x2]
and v5.16b, v4.16b, v1.16b
ushr v4.8h, v4.8h, #8
mul v5.8h, v5.8h, v3.8h
ushr v5.8h, v5.8h, #8
mul v3.8h, v4.8h, v3.8h
bic v3.16b, v3.16b, v1.16b
orr v3.16b, v5.16b, v3.16b
add v2.4s, v2.4s, v3.4s
str q2, [x2]
add x1, x1, #16 // =16
add x2, x2, #16 // =16
sub x0, x0, #4 // =4
b.al #-76
cmp x0, #1 // =1
b.lt #76
ldr s2, [x1]
ldr q3, #72
tbl v3.16b, { v2.16b }, v3.16b
sub v3.8h, v0.8h, v3.8h
ldr s4, [x2]
and v5.16b, v4.16b, v1.16b
ushr v4.8h, v4.8h, #8
mul v5.8h, v5.8h, v3.8h
ushr v5.8h, v5.8h, #8
mul v3.8h, v4.8h, v3.8h
bic v3.16b, v3.16b, v1.16b
orr v3.16b, v5.16b, v3.16b
add v2.4s, v2.4s, v3.4s
str s2, [x2]
add x1, x1, #4 // =4
add x2, x2, #4 // =4
sub x0, x0, #1 // =1
b.al #-76
ret
after: ldr q0, #168
ldr q1, #180
ldr q2, #192
cmp x0, #4 // =4
b.lt #72
ldr q3, [x1]
tbl v4.16b, { v3.16b }, v0.16b
sub v4.8h, v1.8h, v4.8h
ldr q5, [x2]
and v6.16b, v5.16b, v2.16b
ushr v5.8h, v5.8h, #8
mul v6.8h, v6.8h, v4.8h
ushr v6.8h, v6.8h, #8
mul v4.8h, v5.8h, v4.8h
bic v4.16b, v4.16b, v2.16b
orr v4.16b, v6.16b, v4.16b
add v3.4s, v3.4s, v4.4s
str q3, [x2]
add x1, x1, #16 // =16
add x2, x2, #16 // =16
sub x0, x0, #4 // =4
b.al #-72
cmp x0, #1 // =1
b.lt #72
ldr s3, [x1]
tbl v4.16b, { v3.16b }, v0.16b
sub v4.8h, v1.8h, v4.8h
ldr s5, [x2]
and v6.16b, v5.16b, v2.16b
ushr v5.8h, v5.8h, #8
mul v6.8h, v6.8h, v4.8h
ushr v6.8h, v6.8h, #8
mul v4.8h, v5.8h, v4.8h
bic v4.16b, v4.16b, v2.16b
orr v4.16b, v6.16b, v4.16b
add v3.4s, v3.4s, v4.4s
str s3, [x2]
add x1, x1, #4 // =4
add x2, x2, #4 // =4
sub x0, x0, #1 // =1
b.al #-72
ret
Change-Id: I352a98d3ac2ad84c338330ef4cfae0292a0b32da
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/229064
Reviewed-by: Mike Klein <mtklein@google.com>
Commit-Queue: Mike Klein <mtklein@google.com>
This first tries to JIT while hoisting all constants,
and if that fails, tries again hoisting no constants.
I figure this is one of those 80/20 deals for how to
handle constant hoisting and register pressure. This
probably mostly moots doing anything fancy like using
memory operands with AVX or lane operands with NEON.
This _doesn't_ moot hoisting the NEON tbl arguments,
which is not yet done here, but probably my next CL.
Change-Id: Id09d5cdddcdb45207bdfc914a5a3128a481a26f3
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/229058
Reviewed-by: Mike Klein <mtklein@google.com>
Commit-Queue: Mike Klein <mtklein@google.com>
Even if a JIT ultimately doesn't end up hoisting any values, it's going
to want this information while it decides. Writing it in one place also
ensures we only get it wrong in one place...
I'm no_ extending the lifetime of hoisted instructions here in Builder.
That's something to leave to the backend so they have the flexibility of
which of these values to hoist, if any. If they don't hoist, they'll
need to know when the value dies.
Moving this information back here lets the test expectation goldens
reflect the hoist bit again too. Kind of nice.
Change-Id: Ib165ca898a97c1d822cb28fe24f15bae4d570a17
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/229024
Reviewed-by: Mike Klein <mtklein@google.com>
Commit-Queue: Mike Klein <mtklein@google.com>
Change-Id: Iabfc1106fce9926547278ec1335f4888ca86511e
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/229002
Commit-Queue: Michael Ludwig <michaelludwig@google.com>
Reviewed-by: Michael Ludwig <michaelludwig@google.com>
Switch to Dawn's version of dawn_generator.gni. This depends on having
a file called build_overrides/dawn.gni. However, this will also enable
us to use the upstream Dawn BUILD.gn files more easily in the future.
This required adding it to compile.isolate, so the bots can pick it up.
Keeping up with Dawn:
Rename TextureFormat enums.
Rename dawn::BufferUsageBit::TransferDst -> CopyDst.
Removal of GLAD dependency.
SPIRV-Tools update.
Change-Id: Idcd5d1035ed106485dd2503b829e3c3b57a5688b
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/228568
Reviewed-by: Brian Salomon <bsalomon@google.com>
Reviewed-by: Mike Klein <mtklein@google.com>
Commit-Queue: Stephen White <senorblanco@chromium.org>
This will allow chromium to do a dump without crashing so we can get
more information about what's going wrong on webview in the field.
Bug: chromium:977231
Change-Id: I9022921aded735764d36868bb6606674654439bc
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/228389
Reviewed-by: Mike Klein <mtklein@google.com>
Reviewed-by: Khushal Sagar <khushalsagar@chromium.org>
Commit-Queue: Mike Klein <mtklein@google.com>
Auto-Submit: Adrienne Walker <enne@chromium.org>
When Chrome has a LUM16F texture they tell Skia it is R16F. Although this has been working for them so far it causes trouble with some upcoming changes.
Change-Id: I2473f70e4f725128f143c2dfb08adb79f3c7c166
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/228565
Reviewed-by: Brian Salomon <bsalomon@google.com>
We need to sort out an internal compiler error and a crash.
TBR=bsalomon@google.com
Bug: skia:
Change-Id: I8ecccc6ab696fd8c49735ba4690c4ec1f873c15e
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/228936
Reviewed-by: Chris Dalton <csmartdalton@google.com>
Commit-Queue: Chris Dalton <csmartdalton@google.com>
Change-Id: I5c89d5760c16097d658c454950a6632bd427c6ab
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/228637
Reviewed-by: Mike Klein <mtklein@google.com>
Commit-Queue: Mike Klein <mtklein@google.com>
Adds the option to use a multisampled (or mixed sampled) atlas, and
uses the sample mask and stencil buffer instead of coverage counts.
Bug: skia:
Change-Id: I9fb76d17895ae25208124f6c27e37977ac31b5eb
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/227428
Commit-Queue: Chris Dalton <csmartdalton@google.com>
Reviewed-by: Brian Salomon <bsalomon@google.com>
We can always move data around so that an FMA is possible using no more
registers than we would otherwise, and on x86, evne using no more
instructions.
The basic idea here is that if we can't reuse one of the inputs to
destructively host the FMA instruction, the next best thing is to copy
one of the arguments into tmp() and accumulate the FMA there.
Once the FMA has happened, we just need to copy that result to dst().
We can of course skip that copy if dst() == tmp(). On x86 we never need
that copy; dst() and tmp() are picked using the same logic except that
dst may alias one of its inputs, and we only fall into this case after
we've already found it doesn't. So we can just assert dst() == tmp()
rather than check it like we do on ARM.
It's subtle, but I think sound.
I'm using logical-or to copy registers around. This is a little lazy,
but maybe not as lazy as it looks: on ARM that is _the_ way to copy
registers. There's a vmovdqa instruction I could use on x86, TBD.
All paths through this new code were being exercised on ARM, but we
didn't have anything hitting the tmp case on x86, so I've added a new
unit test that hits the corner cases of both implementations.
Change-Id: I5422414fc50c64d491b4933b4b580b784596f291
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/228630
Reviewed-by: Mike Klein <mtklein@google.com>
Commit-Queue: Mike Klein <mtklein@google.com>
Change-Id: I82aec74524f33b3b8ea7592a9a4bf904127b87b6
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/228569
Reviewed-by: Brian Salomon <bsalomon@google.com>
Commit-Queue: Stephen White <senorblanco@chromium.org>
When SkBaseDevice switched to drawGlyphRunList(), we lost the ability to
detect a) constant-Y text and b) default-positioned text.
As a result, the emitted SVG contains lots of redundant/repeating glyph
positions.
This CL enhances SVGTextBuilder to detect and consolidate constant-Y
glyph positions.
Also restore a useful whitespace unit test.
Change-Id: I50568aef1955f75898ebab41441ad5fe418dac43
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/228563
Commit-Queue: Florin Malita <fmalita@chromium.org>
Reviewed-by: Mike Reed <reed@google.com>
We have pack(x,y,imm) = x | (y<<imm) assuming (x & (y<<imm)) == 0.
If we can destroy x, sli (shift-left-insert) lets us implement that
as x |= y << imm. This happens quite often, so you'll see sequences
of pack that used to look like this
shl v4.4s, v2.4s, #8
orr v1.16b, v4.16b, v1.16b
shl v2.4s, v0.4s, #8
orr v0.16b, v2.16b, v3.16b
shl v2.4s, v0.4s, #16
orr v0.16b, v2.16b, v1.16b
now look like this
sli v1.4s, v2.4s, #8
sli v3.4s, v0.4s, #8
sli v1.4s, v3.4s, #16
We can do this thanks to the new simultaneous register assignment
and instruction selection I added. We used to never hit this case.
Change-Id: I75fa3defc1afd38779b3993887ca302a0885c5b1
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/228611
Reviewed-by: Mike Klein <mtklein@google.com>
Commit-Queue: Mike Klein <mtklein@google.com>
Trying to keep most of the structural parts shared between x86_64 and
aarch64. Not sure if this will stay factored like this long-term, but
the last version felt like there was a bit too much redundancy, and I
don't want to write things like register management more often than have
to.
Change-Id: Ieeb21f433715a730c41c85d657c5b33fa4702696
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/228608
Reviewed-by: Mike Klein <mtklein@google.com>
Commit-Queue: Mike Klein <mtklein@google.com>
This matches the logic in GrGLCaps. Note that we only run Vulkan on one
such bot (IntelHD405) and already disabled msaa testing on that bot in
test.py due to the problems it was having.
Change-Id: Ic84a1e8e5eb6deddad5c2bfe93640925a8d6ec0a
Bug: skia:9023
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/228562
Reviewed-by: Brian Salomon <bsalomon@google.com>
Commit-Queue: Brian Osman <brianosman@google.com>
While most callers of decomposeScale only depend on the scale factors
that are returned, image filter decomposition depends on the remaining
matrix as well. Hence, the following necessary updates to work with
the new behavior of decomposeScale:
* Update imagefilter DAG sample to reflect prescaling
* Correct embedded matrix in SkApplyCTMToFilter
* Add comment to clipRectBounds() clarifying coordinate spaces and image filters
But, we want to have decomposeScale using preScale() because it then
produces a remainder matrix that can be used as the transform for the
image filter draw, instead of wrapping the image filter in an
SkMatrixImageFilter as currently done by SkApplyCTMToFilter.
Bug: skia:7211
Change-Id: If14570afb4189cebc75f3815e8ccdde05cb074e1
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/228438
Reviewed-by: Mike Reed <reed@google.com>
Reviewed-by: Robert Phillips <robertphillips@google.com>
Commit-Queue: Michael Ludwig <michaelludwig@google.com>
- get rid of variadic Assembler::byte()... not used very often
- rename Assembler::byte(ptr,n) to bytes()
- align with 0 bytes, get rid of nop()
Change-Id: I7564d3bad00e3f0d1c7a80153c445966914fccf0
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/228601
Reviewed-by: Mike Klein <mtklein@google.com>
Commit-Queue: Mike Klein <mtklein@google.com>
This reverts commit 2f6c8af73a.
Reason for revert: Failing on some bots
Original change's description:
> Add Ganesh support for LUM16F
>
> When Chrome has a LUM16F texture they tell Skia it is R16F. Although this has been working for them so far it causes trouble with some upcoming changes.
>
> Change-Id: Ic2143ec69f33a17cb1cb64f6ebc39bffd94a5e68
> Reviewed-on: https://skia-review.googlesource.com/c/skia/+/228557
> Reviewed-by: Brian Salomon <bsalomon@google.com>
> Commit-Queue: Robert Phillips <robertphillips@google.com>
TBR=bsalomon@google.com,robertphillips@google.com
Change-Id: Ie52150f4f5255a2db697b9f0e810eff2042eadec
No-Presubmit: true
No-Tree-Checks: true
No-Try: true
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/228560
Reviewed-by: Robert Phillips <robertphillips@google.com>
Commit-Queue: Robert Phillips <robertphillips@google.com>
When Chrome has a LUM16F texture they tell Skia it is R16F. Although this has been working for them so far it causes trouble with some upcoming changes.
Change-Id: Ic2143ec69f33a17cb1cb64f6ebc39bffd94a5e68
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/228557
Reviewed-by: Brian Salomon <bsalomon@google.com>
Commit-Queue: Robert Phillips <robertphillips@google.com>
Bug: skia:
Change-Id: I1c801a26727b72f36d76e1a1c21cd0e571107f8c
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/228558
Reviewed-by: Brian Osman <brianosman@google.com>
Commit-Queue: Ethan Nicholas <ethannicholas@google.com>
You can tell how apprehensive I am about this by the number of comments
that I've written, but I think it all makes reasonable sense, and does
mean we can run right up to the line of using all registers, never
wasting a tmp or dst register that would go unused.
I don't think there are any function argument evaluation order issues
here, but it's reassuring that we're testing with GCC and Clang both
when I see things like a->vfoops(dst(), tmp(), r[z]).
Tests pass, and the little big of debug tracing I added temporarily
looked like it made sense. Have not looked at how the disassembly
changes, mostly because I hacked this up on my Mac. Will look before
and after on Linux tomorrow if this sticks.
Change-Id: I1e62aaeba12c07787128ed4a2b67fb8bc27039f6
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/228520
Reviewed-by: Mike Klein <mtklein@google.com>
Commit-Queue: Mike Klein <mtklein@google.com>
Bug: skia:8235
Change-Id: I5a73089c8c9febd3cedf41d0f2e3a52baebda1c9
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/227956
Commit-Queue: Mike Klein <mtklein@google.com>
Reviewed-by: Mike Klein <mtklein@google.com>
This is a reland of 558b639225
PS2... oh, right, not everything supports AVX2.
Original change's description:
> more JIT refactoring
>
> This re-enables AVX2 JIT with simultaneous register assignment and
> instruction selection. You can see it working in a very basic way in
> how we choose instructions and registers for Op::mad_f32.
>
> Constants are still broadcast, here inside the loop instead of hoisted.
> I think it'll probably end up best to use constants directly from memory
> (as in vpshufb's masks), falling back to these in-loop broadcasts when
> that can't work.
>
> Change-Id: If17d51b9960f08da3612e51ac04424e996bf83d4
> Reviewed-on: https://skia-review.googlesource.com/c/skia/+/228366
> Commit-Queue: Mike Klein <mtklein@google.com>
> Reviewed-by: Mike Klein <mtklein@google.com>
Cq-Include-Trybots: skia.primary:Test-Mac10.13-Clang-VMware7.1-CPU-AVX-x86_64-Debug-All-NativeFonts
Change-Id: I6f99d275040abe6210a980fc544f7f22c3b85727
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/228476
Reviewed-by: Mike Klein <mtklein@google.com>
Commit-Queue: Mike Klein <mtklein@google.com>
This reverts commit 558b639225.
Reason for revert: broke some perf bots
Original change's description:
> more JIT refactoring
>
> This re-enables AVX2 JIT with simultaneous register assignment and
> instruction selection. You can see it working in a very basic way in
> how we choose instructions and registers for Op::mad_f32.
>
> Constants are still broadcast, here inside the loop instead of hoisted.
> I think it'll probably end up best to use constants directly from memory
> (as in vpshufb's masks), falling back to these in-loop broadcasts when
> that can't work.
>
> Change-Id: If17d51b9960f08da3612e51ac04424e996bf83d4
> Reviewed-on: https://skia-review.googlesource.com/c/skia/+/228366
> Commit-Queue: Mike Klein <mtklein@google.com>
> Reviewed-by: Mike Klein <mtklein@google.com>
TBR=mtklein@google.com
Change-Id: Id6cd5acd873499bb394009489d77e7636ecbc9c6
No-Presubmit: true
No-Tree-Checks: true
No-Try: true
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/228462
Reviewed-by: Mike Klein <mtklein@google.com>
Commit-Queue: Mike Klein <mtklein@google.com>
This re-enables AVX2 JIT with simultaneous register assignment and
instruction selection. You can see it working in a very basic way in
how we choose instructions and registers for Op::mad_f32.
Constants are still broadcast, here inside the loop instead of hoisted.
I think it'll probably end up best to use constants directly from memory
(as in vpshufb's masks), falling back to these in-loop broadcasts when
that can't work.
Change-Id: If17d51b9960f08da3612e51ac04424e996bf83d4
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/228366
Commit-Queue: Mike Klein <mtklein@google.com>
Reviewed-by: Mike Klein <mtklein@google.com>
This is a reland of f42de9e1e5
Original change's description:
> Interpreter: Bounds check array access, add bool return from run
>
> Out of bounds access with constant indices is a compile error.
> At runtime, causes the interpreter to fail. Made several other
> conditions trigger the same failure logic, and updated all
> uses of the interpreter to validate success.
>
> Change-Id: I3720b3c83903220b010ec574121fc64dbe102378
> Reviewed-on: https://skia-review.googlesource.com/c/skia/+/228256
> Commit-Queue: Brian Osman <brianosman@google.com>
> Reviewed-by: Mike Reed <reed@google.com>
Change-Id: I8849de815f7efb730ac9c55b6edd296cb9ca7599
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/228353
Commit-Queue: Brian Osman <brianosman@google.com>
Reviewed-by: Brian Osman <brianosman@google.com>
Changes the GrBackendSurface to not take it separately.
Change-Id: I5eb9684da1532907e7833e9d7835a74d1e71f5ca
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/228396
Reviewed-by: Greg Daniel <egdaniel@google.com>
Commit-Queue: Brian Salomon <bsalomon@google.com>
With this change I also removed GrPixelConfig as param to supportedReadPixelsColorType().
This meant some updates had to be made to Vulkan and Metal to make sure they return the
right GrColorType.
Bug: skia:6718
Change-Id: I71b6360489cf499692c7b777e5915090fad05c56
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/228349
Reviewed-by: Brian Salomon <bsalomon@google.com>
Commit-Queue: Greg Daniel <egdaniel@google.com>
Change-Id: Id1a9830bdead5ed7cd52c61c31298d8279c337b0
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/228136
Commit-Queue: Stephen White <senorblanco@chromium.org>
Reviewed-by: Brian Salomon <bsalomon@google.com>
Change-Id: Ic30f6e0345e851ea8a942996b9eaf2c894455e3e
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/228236
Reviewed-by: Brian Osman <brianosman@google.com>
Commit-Queue: Mike Reed <reed@google.com>
I find myself passing around parallel vectors of Builder::Instructions
and deaths so often that it just makes more sense practically to store
them together. It's a little awkward that the values are only useful
after calling done(), but I can live with that.
Get a little more careful about mutation, passing Builder::Instructions
by const&. Instead of extending lifetimes of live hoisted
instructions, just check for them in maybe_recycle_register() instead.
Change-Id: I1cb9e25c1a7c46a250c2271334821be8535353bf
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/228367
Reviewed-by: Mike Klein <mtklein@google.com>
Commit-Queue: Mike Klein <mtklein@google.com>
Add a test case that previously asserted (due to incorrect code-gen
underflowing the stack).
Change-Id: I9df7a08b9ac5c7b5bc246129f3383dc723173351
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/228351
Reviewed-by: Ethan Nicholas <ethannicholas@google.com>
Commit-Queue: Brian Osman <brianosman@google.com>
First draft of (mostly stubbed-out) GrDawnGpu.
Skeletons of GrDawnCaps, GrDawnGpuCommandBuffer, GrDawnRenderTarget.
First draft of DawnTestContext.
First draft of psuedo-fences for Dawn, implemented with MapReadAsync.
Change-Id: I443f3370522639e82f2fa0eebe6b206c372f13a4
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/228137
Commit-Queue: Stephen White <senorblanco@chromium.org>
Reviewed-by: Brian Salomon <bsalomon@google.com>
This reverts commit f42de9e1e5.
Reason for revert: All the SANs
Original change's description:
> Interpreter: Bounds check array access, add bool return from run
>
> Out of bounds access with constant indices is a compile error.
> At runtime, causes the interpreter to fail. Made several other
> conditions trigger the same failure logic, and updated all
> uses of the interpreter to validate success.
>
> Change-Id: I3720b3c83903220b010ec574121fc64dbe102378
> Reviewed-on: https://skia-review.googlesource.com/c/skia/+/228256
> Commit-Queue: Brian Osman <brianosman@google.com>
> Reviewed-by: Mike Reed <reed@google.com>
TBR=mtklein@google.com,brianosman@google.com,ethannicholas@google.com,reed@google.com
Change-Id: I434601960d54fbd7d00e2af2dc6269a83a768c5b
No-Presubmit: true
No-Tree-Checks: true
No-Try: true
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/228352
Reviewed-by: Brian Osman <brianosman@google.com>
Commit-Queue: Brian Osman <brianosman@google.com>
Out of bounds access with constant indices is a compile error.
At runtime, causes the interpreter to fail. Made several other
conditions trigger the same failure logic, and updated all
uses of the interpreter to validate success.
Change-Id: I3720b3c83903220b010ec574121fc64dbe102378
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/228256
Commit-Queue: Brian Osman <brianosman@google.com>
Reviewed-by: Mike Reed <reed@google.com>
This reverts commit 9725638fb1.
Reason for revert: Chrome roll
Original change's description:
> Make rest of GrGpu::wrapBackend* methods take a GrColorType
>
> This CL is intended to further wean Ganesh off of using the GrBackendTexture's pixel config
>
> Bug: skia:6718
> Change-Id: I593c0c73922fb76045e379214e20adb1f17ea215
> Reviewed-on: https://skia-review.googlesource.com/c/skia/+/227780
> Commit-Queue: Robert Phillips <robertphillips@google.com>
> Reviewed-by: Greg Daniel <egdaniel@google.com>
TBR=egdaniel@google.com,bsalomon@google.com,robertphillips@google.com
Change-Id: Id71acf1dec63c288a858fccd7109c84cf3cc6f0a
No-Presubmit: true
No-Tree-Checks: true
No-Try: true
Bug: skia:6718
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/228337
Reviewed-by: Robert Phillips <robertphillips@google.com>
Commit-Queue: Robert Phillips <robertphillips@google.com>
Change-Id: I04493af0a6ce1425c4acf68365135722dd3c218b
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/227857
Commit-Queue: Michael Ludwig <michaelludwig@google.com>
Reviewed-by: Robert Phillips <robertphillips@google.com>
Do less in Builder, more in Program::Program().
This temporarily disables JITs until I rewrite
them to also build from these new inputs:
- vector<Builder::Instruction>
- vector<Val> deaths
- vector<int> strides
i.e. to do their own register assignment
and make their own hoisting decisions.
Change-Id: Ie2ce9755f20860a80506e913b7b139d562e291c3
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/228216
Reviewed-by: Mike Klein <mtklein@google.com>
Commit-Queue: Mike Klein <mtklein@google.com>
This CL is intended to further wean Ganesh off of using the GrBackendTexture's pixel config
Bug: skia:6718
Change-Id: I593c0c73922fb76045e379214e20adb1f17ea215
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/227780
Commit-Queue: Robert Phillips <robertphillips@google.com>
Reviewed-by: Greg Daniel <egdaniel@google.com>
I'm slowly refactoring my way to where hoisting and register assignment
are done in backend-specific ways, but this liveness analysis is always
going to be useful for each backend.
Use deaths() to restore friendly ☠️ dead code markers in test dumps.
Change-Id: I3ab94665bbbbf0788b0b27e00d644eba927dff47
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/228113
Commit-Queue: Mike Klein <mtklein@google.com>
Reviewed-by: Mike Klein <mtklein@google.com>
Change-Id: Ie88b935a75d8b9c253b29bf0aa4ef39649327bfa
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/228119
Reviewed-by: Brian Osman <brianosman@google.com>
Commit-Queue: Mike Reed <reed@google.com>
This reverts commit 6cebea42a8.
Reason for revert: breaking google3 roll
Original change's description:
> First draft of Dawn backend: clears are working.
>
> First draft of (mostly stubbed-out) GrDawnGpu.
> Skeletons of GrDawnCaps, GrDawnGpuCommandBuffer, GrDawnRenderTarget.
> First draft of DawnTestContext.
> First draft of psuedo-fences for Dawn, implemented with MapReadAsync.
>
> Change-Id: Id009436f4441f26ffbc82d485d7af3a499b3281b
> Reviewed-on: https://skia-review.googlesource.com/c/skia/+/226857
> Commit-Queue: Stephen White <senorblanco@chromium.org>
> Reviewed-by: Greg Daniel <egdaniel@google.com>
> Reviewed-by: Brian Salomon <bsalomon@google.com>
TBR=egdaniel@google.com,jvanverth@google.com,bsalomon@google.com,senorblanco@chromium.org
Change-Id: Ie494b5a403e8537c6539551533ae8b9156e90a61
No-Presubmit: true
No-Tree-Checks: true
No-Try: true
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/228120
Reviewed-by: Greg Daniel <egdaniel@google.com>
Commit-Queue: Greg Daniel <egdaniel@google.com>
Change-Id: I7fd3a8f6c02217d011f353ef602718a537bb87b2
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/228116
Reviewed-by: Brian Osman <brianosman@google.com>
Commit-Queue: Greg Daniel <egdaniel@google.com>
First draft of (mostly stubbed-out) GrDawnGpu.
Skeletons of GrDawnCaps, GrDawnGpuCommandBuffer, GrDawnRenderTarget.
First draft of DawnTestContext.
First draft of psuedo-fences for Dawn, implemented with MapReadAsync.
Change-Id: Id009436f4441f26ffbc82d485d7af3a499b3281b
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/226857
Commit-Queue: Stephen White <senorblanco@chromium.org>
Reviewed-by: Greg Daniel <egdaniel@google.com>
Reviewed-by: Brian Salomon <bsalomon@google.com>
'void main(float x, float y, inout half4 color)'
For colorfilters, x,y are uninitialized
For shaders, color is paint-color
Change-Id: I21c2f570702f117d33646bd8fba0ab4def20a118
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/227439
Reviewed-by: Brian Osman <brianosman@google.com>
Commit-Queue: Mike Reed <reed@google.com>
Bug: skia:
Change-Id: I222cd0afad2c5226f8ccdec5c5c989c56671c2ff
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/228057
Reviewed-by: Brian Osman <brianosman@google.com>
Commit-Queue: Ethan Nicholas <ethannicholas@google.com>
This is the last surface desc flag, so remove flags from GrSurfaceDesc.
Bug: skia:6718
Change-Id: Id3ed66b161289927b62f40bfb1f6482cf544deda
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/227858
Reviewed-by: Robert Phillips <robertphillips@google.com>
Reviewed-by: Greg Daniel <egdaniel@google.com>
Commit-Queue: Brian Salomon <bsalomon@google.com>
This has failed in the past on GPU perf bots,
I think because I've been flushing the cache wrong?
Try switching over to __builtin___clear_cache(), what I used
a few years ago when I last tried to do something like this.
Cq-Include-Trybots: skia.primary:Perf-Android-Clang-Nexus5x-GPU-Adreno418-arm64-Release-All-Android
Change-Id: Ic5e366022eb15f4544fb7b90a75f8c51e36392ca
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/227421
Commit-Queue: Mike Klein <mtklein@google.com>
Reviewed-by: Herb Derby <herb@google.com>
Add Program::dropJIT() to allow us to proactively drop
any JIT code forcing fallback on the interpreter,
and use it to test both on JIT-supported platforms.
Other platforms will just test the interpreter twice.
Change-Id: I607d00ef3c648e66a0b3a1374b11aa82dbfff70c
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/227424
Commit-Queue: Mike Klein <mtklein@google.com>
Reviewed-by: Herb Derby <herb@google.com>
In the new world if the GrColorType cannot creating a support backend format
it returns an invalid GrBackendFormat. So we need to check this in our
fallback when uploading raster image to gpu.
Change-Id: Ie69a06ba1a0badb5b7e78965906f09c0413453e2
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/227859
Reviewed-by: Brian Salomon <bsalomon@google.com>
Commit-Queue: Greg Daniel <egdaniel@google.com>
Move JITting from lazy in eval() to proactive in Program::Program().
There's no need to delay to eval() now that strides are known up front.
There's _still_ one more reason we need to keep the interpreter around
even if we can JIT... can_jit() may return false (too many regs, too
many args).
Change-Id: I0a176b97bcd9e8d0fcf2a9fa4b7f64103fd51e75
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/227419
Commit-Queue: Mike Klein <mtklein@google.com>
Reviewed-by: Herb Derby <herb@google.com>
Arg strides are the reason JIT happens lazily in Program::eval() today
instead of proactively in Builder::done() or Program's constructor. It
also just really doesn't make sense to delay this information... it's
not like you can change it up sanely between calls to eval().
The argument index now comes implicitly from the order of calling arg().
This may seem logically independent, but it prevents a weird situation
where you could use the same argument index twice with different
strides... not sure what that would mean.
Change-Id: I0f5d46e94a1ca112a72675c5492f17c0dd825ce0
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/227390
Reviewed-by: Herb Derby <herb@google.com>
Commit-Queue: Mike Klein <mtklein@google.com>
Removes some usage of GrSurfaceDesc.
Bug: skia:6718
Change-Id: Icc4f93aba0e5c49a801b4c7bbfcba76a6e30c538
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/227776
Commit-Queue: Brian Salomon <bsalomon@google.com>
Reviewed-by: Robert Phillips <robertphillips@google.com>
This reverts commit 8372e4bb40.
Reason for revert: This doesn't really work while RT has both resolve and msaa in it since chrome cant blit with different formats and we can't check that during a resolve correctly.
Original change's description:
> Changes to gl BGRA renderbuffer format support.
>
> Change-Id: Ic46057ed1f85e368fd55148beaa3f7e2ae54a042
> Reviewed-on: https://skia-review.googlesource.com/c/skia/+/227398
> Commit-Queue: Greg Daniel <egdaniel@google.com>
> Reviewed-by: Brian Salomon <bsalomon@google.com>
TBR=egdaniel@google.com,bsalomon@google.com
Change-Id: Ia28b8ef1e5f4274240ad207e34619049525bd722
No-Presubmit: true
No-Tree-Checks: true
No-Try: true
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/227781
Reviewed-by: Greg Daniel <egdaniel@google.com>
Commit-Queue: Greg Daniel <egdaniel@google.com>
This mirrors the logic in emitCode, and avoids extra shader compiles.
Change-Id: I2c11e082dfa2b9fc673dc2b8fdf2f0b964407b90
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/227777
Commit-Queue: Brian Osman <brianosman@google.com>
Commit-Queue: Brian Salomon <bsalomon@google.com>
Auto-Submit: Brian Osman <brianosman@google.com>
Reviewed-by: Brian Salomon <bsalomon@google.com>
1) It only applies when a texture is created, not when recycled from cache
2) It is all textures or none, not a flag GrSurfaceDesc
3) It is implemented by GrGpu clearing the texture after creation if
such a thing is supported in underlying API. Otherwise, GrResourceProvider
must provide pre-zeroed mip levels.
4) Works for MIP mapped textures (all levels without initial data are cleared)
This could cause performance regressions in WebGL until we re-add the
ability to clear using glCear() in GL. Doing that requires making the "can
clear using GrGpu" caps query be per-format. Deferring doing that until
GrPixelConfig work is farther along.
Bug: skia:6718
Change-Id: I234715b9faaf61e8b44d54464497a17cd553585d
start
Change-Id: Ib84a8c3ece010cc3164b18895107e78484cbf76b
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/226977
Reviewed-by: Robert Phillips <robertphillips@google.com>
Commit-Queue: Brian Salomon <bsalomon@google.com>
This code checks the external format/type that will be used with
glReadPixels. It tests for the values inherently allowed by the GL
and for implementation values.
It would need to be refactored to use formats and color types rather
than GrPixelConfig.
However, the code that calls GrGpu::readPixels and
GrGpu::transferPixelsFrom already tests
GrCaps::supportedReadPixelsColorType. The GrGLCaps override only allows
format/type combinations allowed by ES without implementation queries,
which is less permissive than the deleted code.
We may get more permissive again but will probably follow a different
pattern and do this after the format/colortype tables are built out in
GrGLCaps.
Bug: skia:6718
Change-Id: I36d9ab496dfe71045dcbce84200a9de3e93440bf
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/226840
Commit-Queue: Brian Salomon <bsalomon@google.com>
Reviewed-by: Greg Daniel <egdaniel@google.com>
Make them query caps for supported read/write info and do CPU
conversions before uploading/after reading.
Removes use of GrColor so in theory could be used to test
non-8888 color types.
Bug: skia:6718
Change-Id: Icf9d0b778348a4e960fbfec49e1308b21e45a051
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/227497
Reviewed-by: Greg Daniel <egdaniel@google.com>
Commit-Queue: Brian Salomon <bsalomon@google.com>
Change-Id: I54a4913e5ec2e9ed41d754c1f0cd2e6d7bbeb7a8
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/227640
Reviewed-by: Brian Osman <brianosman@google.com>
Commit-Queue: Brian Osman <brianosman@google.com>
This is a step towards reducing our reliance-on/use-of the GrPixelConfig stored in the GrBackendTexture.
TBR=egdaniel@google.com
Bug: skia:6718
Change-Id: I316a98416c51f273e6ab578f9cbaea5f7adfe331
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/227639
Reviewed-by: Robert Phillips <robertphillips@google.com>
Commit-Queue: Robert Phillips <robertphillips@google.com>
This reverts commit 5572737d95.
Reason for revert: Adreno 4xx bots hitting compatibility assert in InitialTextureClear test
Original change's description:
> Pass GrColorType to the GrGpu::wrapRenderableBackendTexture chain of calls
>
> This is a step towards reducing our reliance-on/use-of the GrPixelConfig stored in the GrBackendTexture.
>
> Bug: skia:6718
> Change-Id: I2170032bfbbb57423c4bb0d901ad014c61d38131
> Reviewed-on: https://skia-review.googlesource.com/c/skia/+/223701
> Commit-Queue: Robert Phillips <robertphillips@google.com>
> Reviewed-by: Greg Daniel <egdaniel@google.com>
TBR=egdaniel@google.com,robertphillips@google.com
Change-Id: I24cf6b3aa0dfca2e935a36592860ad91171b21a7
No-Presubmit: true
No-Tree-Checks: true
No-Try: true
Bug: skia:6718
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/227637
Reviewed-by: Greg Daniel <egdaniel@google.com>
Commit-Queue: Greg Daniel <egdaniel@google.com>
This is a step towards reducing our reliance-on/use-of the GrPixelConfig stored in the GrBackendTexture.
Bug: skia:6718
Change-Id: I2170032bfbbb57423c4bb0d901ad014c61d38131
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/223701
Commit-Queue: Robert Phillips <robertphillips@google.com>
Reviewed-by: Greg Daniel <egdaniel@google.com>
Change-Id: If581c8ceeaa76985535cb7b6772742f0011cfe8e
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/227436
Commit-Queue: Hal Canary <halcanary@google.com>
Commit-Queue: Ben Wagner <bungeman@google.com>
Auto-Submit: Hal Canary <halcanary@google.com>
Reviewed-by: Ben Wagner <bungeman@google.com>
Change-Id: I171a680aac554a0015d1854c46b35e9c9785fdf3
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/227061
Reviewed-by: Mike Klein <mtklein@google.com>
Commit-Queue: Brian Osman <brianosman@google.com>
Previously we would just run coverage count transformations on
everything, including cached literal coverage atlases. This was
wasteful since it isn't necessary if the atlas already has literal
coverage. MSAA mode will introduce even more atlases that don't need
coverage count transformations, so it's definitely time to clean this
up.
Bug: skia:
Change-Id: Ifc72eaa7cbd4ab5e4ef4acb5610117ae9f54e4c1
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/227144
Reviewed-by: Brian Salomon <bsalomon@google.com>
Commit-Queue: Chris Dalton <csmartdalton@google.com>
Basically the same deal as aarch64:
- a bunch of instructions to rewrite control
flow to be two loops, body and tail
- a bunch of instructions to support scalar
loads and stores in the tail
We can now remove the JIT::mask field.
I've removed the SkUNREACHABLE I'd put in for the ARM code... as
written the interpreter is still reachable by the loser if two threads
race to JIT the program. Medium term I plan to move JIT compilation to
a more proactive time, eliminating the need for the lock and letting the
interpreter become truly unreachable.
I had a little bit of a false start with what instructions to use for
scalar load8 and store8, first starting with instructions that loaded
via GP registers, then remembering vpinsrb and vpextrb can take a memory
argument, loading into xmm directly. I've left the first instructions I
used in the file, still implemented but only used from the unit tests.
They're pretty common and will probably be useful some day.
Change-Id: I471b13026af4b1c6e861a53159f9df5f0285447c
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/227178
Reviewed-by: Herb Derby <herb@google.com>
Commit-Queue: Mike Klein <mtklein@google.com>
I had been setting the REX R bit to select high registers,
but you actually set the B bit. Don't know how I got that
wrong before... the leading byte should be 49 not 4c.
$ cat test.s
foo:
addq $7, %r8
$ clang -c test.s && objdump -d test.o
0000000000000000 <foo>:
0: 49 83 c0 07 add $0x7,%r8
Change-Id: I039e1c4f4ea20523a1e2cc9bcf5f6d9321a6223b
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/227177
Reviewed-by: Herb Derby <herb@google.com>
Commit-Queue: Mike Klein <mtklein@google.com>
This reverts commit 8c5c54ceaf.
Reason for revert: bad ANGLE diffs, Chrome roll failing
Original change's description:
> Use saveLayer-determined CT in SkGpuDevice::onCreateDevice
>
> Currently Ganesh always uses the prev render target CT, but saveLayer
> may force a different CT, depending on flags.
>
> This fixes handling of saveLayer/kF16ColorType (added in
> https://skia-review.googlesource.com/c/skia/+/227066).
>
> Change-Id: I8e73fdfdf18eb61f97fadd0504a4bd0f4a71977d
> Reviewed-on: https://skia-review.googlesource.com/c/skia/+/227258
> Reviewed-by: Brian Salomon <bsalomon@google.com>
> Commit-Queue: Florin Malita <fmalita@chromium.org>
TBR=bsalomon@google.com,fmalita@chromium.org
Change-Id: I0c3e74be64d5286b8d95b2aabaa464ad97fcaccf
No-Presubmit: true
No-Tree-Checks: true
No-Try: true
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/227262
Reviewed-by: Florin Malita <fmalita@google.com>
Commit-Queue: Florin Malita <fmalita@google.com>
Currently Ganesh always uses the prev render target CT, but saveLayer
may force a different CT, depending on flags.
This fixes handling of saveLayer/kF16ColorType (added in
https://skia-review.googlesource.com/c/skia/+/227066).
Change-Id: I8e73fdfdf18eb61f97fadd0504a4bd0f4a71977d
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/227258
Reviewed-by: Brian Salomon <bsalomon@google.com>
Commit-Queue: Florin Malita <fmalita@chromium.org>
Add a tail loop to handle elements one at a time.
Just like in the interpreter, the only instructions
that need to be changed are the loads and stores,
16 byte -> 4 byte and 4 byte -> 1 byte.
With this we can mark the interpreter as SkUNREACHABLE,
and it even completely compiles away, saving a few KB.
Example profile for the SkVMTool float-squaring program
running N=15 over and over:
Samples│
│ skvm-jit-3663518994():
42 │40: cmp x0, #0x4
│44: ↓ b.lt 60
51 │48: ldr q0, [x1]
197 │4c: mul v0.4s, v0.4s, v0.4s
135 │50: str q0, [x1]
│54: add x1, x1, #0x10
43 │58: sub x0, x0, #0x4
│5c: b.al 40
150 │60: ↓ cbz x0, 7c
67 │64: ldr s0, [x1]
130 │68: mul v0.4s, v0.4s, v0.4s
135 │6c: str s0, [x1]
18 │70: add x1, x1, #0x4
17 │74: sub x0, x0, #0x1
20 │78: b.al 60
124 │7c: ← ret
Change-Id: I153d7bc247942366a686e30a9cad60c935f754ed
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/227138
Commit-Queue: Mike Klein <mtklein@google.com>
Reviewed-by: Herb Derby <herb@google.com>
These didn't work correctly, and they're extremely tricky to get right
in the vectorized execution model (vs. structured control flow). As a
side effect, determine the maximum stack depth used for the execution
masking - the same idea will be used for the primary stack in a later
CL. Add a unit test to verify the new restriction, and fix two places
that were relying on this feature before.
In addition, boolean external values need to be masks. I may implement
this in the code-gen at some point, but this is already a fringe
feature, so just fix the one unit test for now.
Change-Id: I9607ffaf67c7795dbf42e4009180aea8f3e65c44
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/226849
Reviewed-by: Mike Klein <mtklein@google.com>
Commit-Queue: Brian Osman <brianosman@google.com>
This turned out to be quite an easy transformation
with yesterday's work already done. No codegen changes.
Change-Id: Ife19ab7731514c54cfed963a6d2e9b1ec2246997
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/227137
Commit-Queue: Mike Klein <mtklein@google.com>
Reviewed-by: Herb Derby <herb@google.com>
Also rename from GrGLSizedInternalFormat since the compressed formats
aren't really sized internal formats.
Change-Id: I8744f1f4b8156300ab69d89066b44913bce120a3
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/226956
Reviewed-by: Greg Daniel <egdaniel@google.com>
Commit-Queue: Brian Salomon <bsalomon@google.com>
Found by a GCC warning that
fPhase = that.fPhase; // line 193
was sometimes copying over uninitialized data.
Seems pretty harmless to initialize.
Change-Id: I9011ccf965bea4a702e0cca61f89e0aa644c55a6
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/227122
Commit-Queue: Mike Klein <mtklein@google.com>
Reviewed-by: Brian Salomon <bsalomon@google.com>
We were checking 'externalType' twice, and never checking
'internalFormat'. We now check each of them once.
Change-Id: Iae0578a4828fd2a9224e4924245bea0708d8c931
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/227028
Reviewed-by: Brian Salomon <bsalomon@google.com>
Commit-Queue: Brian Salomon <bsalomon@google.com>
This reverts commit aed8009a6d.
Reason for revert: Flutter's version of Wuffs has been updated. See https://github.com/flutter/engine/pull/9466#issuecomment-510639898
Original change's description:
> Revert "Update Wuffs version"
>
> This reverts commit 42ece2b7c9.
>
> Reason for revert: Requiring the latest version of wuffs broke the flutter roll.
>
> Original change's description:
> > Update Wuffs version
> >
> > The primary purpose of this commit is to track upstream Wuffs more
> > closely.
> >
> > A side effect is to pull in the Wuffs commit
> > 5bea867f72
> > "Allow an LZW literal width of 1", which eliminates a difference between
> > the old third_party/gif decoder and the new third_party/wuffs decoder.
> >
> > As the CodecTest.cpp comment says, the GIF spec explicitly says that the
> > LZW literal width should be at least 2, but in practice, GIF encoders
> > violate the spec. After that upstream commit, Wuffs has followed other
> > GIF decoders in being more liberal in what it accepts.
> >
> > Codec_InvalidAnimated therefore no longer has a separate "#ifdef
> > SK_HAS_WUFFS_LIBRARY" section. The first frame of the test's GIF image
> > data, being the required frame of the third frame, no longer has an
> > invalid LZW literal width according to Wuffs.
> >
> > Bug: skia:8235
> > Change-Id: Ie94537f5232128ffc1d1547f4c0b84992e54ab02
> > Reviewed-on: https://skia-review.googlesource.com/c/skia/+/226476
> > Commit-Queue: Leon Scroggins <scroggo@google.com>
> > Reviewed-by: Leon Scroggins <scroggo@google.com>
>
> TBR=scroggo@google.com,nigeltao@google.com
>
> Change-Id: I9e636e81f57eefd836a53738872ddb9f5c9b13c3
> No-Presubmit: true
> No-Tree-Checks: true
> No-Try: true
> Bug: skia:8235
> Reviewed-on: https://skia-review.googlesource.com/c/skia/+/226697
> Reviewed-by: Leon Scroggins <scroggo@google.com>
> Commit-Queue: Leon Scroggins <scroggo@google.com>
TBR=scroggo@google.com,nigeltao@google.com
Change-Id: Ibeeea1cf9c2e210b5e49dec65037ec8a494209de
No-Presubmit: true
No-Tree-Checks: true
No-Try: true
Bug: skia:8235
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/226851
Reviewed-by: Leon Scroggins <scroggo@google.com>
Reviewed-by: Nigel Tao <nigeltao@google.com>
Commit-Queue: Leon Scroggins <scroggo@google.com>
Support forward references in Label.
In addition to tracking the current Label offset (used for
backward references essentially just the same as before this CL)
we also store a list of instructions that refer to each Label.
When a Label moves, each instruction gets a new displacement.
To make this a little easier, remove the 8-bit jump form on x86...
this way all x86 displacements are 32-bit and and all ARM 19-bit.
For now only cbz() supports this, just to start somewhere.
More to do but it's worth an early design review.
Change-Id: I23d2bcd7742965ab694ae4828f53409cb9fc807f
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/226937
Commit-Queue: Mike Klein <mtklein@google.com>
Reviewed-by: Herb Derby <herb@google.com>
Adds a separate resolve texture to GrMtlRenderTarget, which can be used
to do a resolve for the main multisample color texture. The resolve is
handled by setting a special Store action for the RenderCommandEncoder.
Bug: skia:8243
Change-Id: I1ffd756c01a9b363116ffefee2c4c50ba9a3e637
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/225536
Reviewed-by: Greg Daniel <egdaniel@google.com>
Commit-Queue: Jim Van Verth <jvanverth@google.com>
Also adds the following to the format table:
*base internal format
*sized internal format
*compressed internal format
*the internal format to use with glTexImage
*the external type to use with glTexImage when there is no data to upload
(i.e., no GrColorType for data)
Bug: skia:6718
Change-Id: Ica51a8d4588bd24078c7d61805f6eef9c02ae14d
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/226558
Commit-Queue: Brian Salomon <bsalomon@google.com>
Reviewed-by: Greg Daniel <egdaniel@google.com>
MakeFromBackendTextureAsRenderTarget is planned to be deprecated, so we
should use MakeFromBackendTexture with a sampleCount parameter instead.
On Vulkan, this ran into issues because we assumed an allocation for the
VkImage and the swapchain doesn't provide us with one. Fixed so we don't
need an allocation for Borrowed textures.
Bug: skia:
Change-Id: Ib26888020e093f4a734a4159eae898539c2273b7
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/226839
Reviewed-by: Greg Daniel <egdaniel@google.com>
Commit-Queue: Jim Van Verth <jvanverth@google.com>
Change-Id: I231949c90342a44e9c0b6030818139d3164819da
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/226837
Reviewed-by: Brian Salomon <bsalomon@google.com>
Commit-Queue: Jim Van Verth <jvanverth@google.com>
This adds a bunch of instructions we'll need to handle the N < 4 tail
within the JIT code on ARM.
- ldrb/strb are 1-byte load and stores
- sub subtracts without setting flags
- cmp just sets flags (actually just subs with an xzr destination)
- add b and b.lt, just like b.ne
- cbz and cbnz... we only need cbz but I accidentally did cbnz first
Once I add support for forward jumps, we'll be able to use these
instructions to restructure the loop to
entry:
hoisted setup
loop:
if N < 4, jump tail (cmp N,#4; b.lt tail)
... handle 4 values ...
jump loop (b loop)
tail:
if N == 0, jump end (cbz N, end)
... handle 1 value ...
jump tail (b tail)
end:
ret
Change-Id: I62d2d190f670f758197a25d99dfde13362189993
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/226828
Reviewed-by: Herb Derby <herb@google.com>
Commit-Queue: Mike Klein <mtklein@google.com>
We conflate luminance (dot of linear r,g,b) and luma (dot of
gamma-encoded r',g',b') all over the place in Skia. This one is
possibly the most confusing of any of them, in that the dot-product
coefficients for BT.709 luma were somewhat arbitrarily chosen to be the
same as the coefficients for luminance.
So, big old ridiculously new long name to make it clear it works
for both luma and luminance, but that it's hardcoded to BT.709.
My quick reference when I forget things:
https://en.wikipedia.org/wiki/Luma_(video)
Change-Id: I5a6567de296795f558acc5dd3c39974b8035234e
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/226762
Reviewed-by: Brian Osman <brianosman@google.com>
Commit-Queue: Mike Klein <mtklein@google.com>
As the powers-of-2 get larger the coarse binning can burn a lot of VRAM.
Granted it isn't the best metric but, with this CL, the number of textures created and scratch textures reused remains unchanged when running the GMs.
Change-Id: I84abbbae0ed01aabb387671b5ee0e4fcdb82b671
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/226226
Commit-Queue: Robert Phillips <robertphillips@google.com>
Reviewed-by: Brian Salomon <bsalomon@google.com>
This reverts commit 42ece2b7c9.
Reason for revert: Requiring the latest version of wuffs broke the flutter roll.
Original change's description:
> Update Wuffs version
>
> The primary purpose of this commit is to track upstream Wuffs more
> closely.
>
> A side effect is to pull in the Wuffs commit
> 5bea867f72
> "Allow an LZW literal width of 1", which eliminates a difference between
> the old third_party/gif decoder and the new third_party/wuffs decoder.
>
> As the CodecTest.cpp comment says, the GIF spec explicitly says that the
> LZW literal width should be at least 2, but in practice, GIF encoders
> violate the spec. After that upstream commit, Wuffs has followed other
> GIF decoders in being more liberal in what it accepts.
>
> Codec_InvalidAnimated therefore no longer has a separate "#ifdef
> SK_HAS_WUFFS_LIBRARY" section. The first frame of the test's GIF image
> data, being the required frame of the third frame, no longer has an
> invalid LZW literal width according to Wuffs.
>
> Bug: skia:8235
> Change-Id: Ie94537f5232128ffc1d1547f4c0b84992e54ab02
> Reviewed-on: https://skia-review.googlesource.com/c/skia/+/226476
> Commit-Queue: Leon Scroggins <scroggo@google.com>
> Reviewed-by: Leon Scroggins <scroggo@google.com>
TBR=scroggo@google.com,nigeltao@google.com
Change-Id: I9e636e81f57eefd836a53738872ddb9f5c9b13c3
No-Presubmit: true
No-Tree-Checks: true
No-Try: true
Bug: skia:8235
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/226697
Reviewed-by: Leon Scroggins <scroggo@google.com>
Commit-Queue: Leon Scroggins <scroggo@google.com>
Fix: Use non-zero rowbytes when uploading zeros for initial texture clear
This reverts commit f16020ba1b.
Bug: chromium:981254
Change-Id: Iafd5893dd1b397ec1d91c64d48d46059e62488a7
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/226557
Reviewed-by: Greg Daniel <egdaniel@google.com>
Commit-Queue: Brian Salomon <bsalomon@google.com>
Bug: b/137017906
Change-Id: I98bfc9607fa78536e1cf44707636558ed1980b4a
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/226616
Commit-Queue: Greg Daniel <egdaniel@google.com>
Commit-Queue: Brian Osman <brianosman@google.com>
Auto-Submit: Greg Daniel <egdaniel@google.com>
Reviewed-by: Brian Osman <brianosman@google.com>
This is basically a no-op CL, with two changes designed to make
SkVM lifetime analysis easier to reason about:
1) rename Analysis.life to Analysis.death,
since it indicates the instruction when the value dies
2) use death == 0 to mark dead code instead of an NA sentinel
The life->death change really is just a naming thing, something I
realized makes a little more sense this way around after talking about
how this code works out loud. 'death' holds the time of a Val's death.
The second change also isn't very important, but I realized that there
is a perfectly good in-band value to mark dead code, 0, and there's no
need to use NA as a sentinel. If we mark an instruction as death == 0,
that indicates the value is needed only until instruction 0 executes,
i.e. never. They're sort of pre-dead. This cuts one of the overloads
for what NA means in SkVM.cpp; now it only means "no value".
All the code that tested against the NA sentinel now tests against 0
instead. We could go a step further and rewrite the tests to be death >
id for live code or death < id for dead code, but that amounts to
roughly the same thing in the end: instructions either live for some
time with a death that is later than their own ID, or are dead with
death == 0. There is some small ambiguity around whether we should mark
store instruction's deaths as id or id+1 and which of the tests should
be <= or >=, but checking against 0 makes that all moot, and I think the
checks also stand out more clearly with the literal '0' in the code.
This is a little warmup to refamiliarize myself with the code, with an
ultimate goal of moving hoisting and register assignment to the backends
so they can be influenced by instruction selection: platform specific
ways to handle immediates or splats, choosing destination registers and
instructions that play well with the available argument registers, etc.
Change-Id: I6978abf0bd01dcd0e7a142d632826e7692060ade
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/226549
Reviewed-by: Herb Derby <herb@google.com>
Commit-Queue: Mike Klein <mtklein@google.com>
Change-Id: I23a88524d741cecefa16ae7e364d2294db1c6030
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/226508
Reviewed-by: Jim Van Verth <jvanverth@google.com>
Commit-Queue: Robert Phillips <robertphillips@google.com>
The primary purpose of this commit is to track upstream Wuffs more
closely.
A side effect is to pull in the Wuffs commit
5bea867f72
"Allow an LZW literal width of 1", which eliminates a difference between
the old third_party/gif decoder and the new third_party/wuffs decoder.
As the CodecTest.cpp comment says, the GIF spec explicitly says that the
LZW literal width should be at least 2, but in practice, GIF encoders
violate the spec. After that upstream commit, Wuffs has followed other
GIF decoders in being more liberal in what it accepts.
Codec_InvalidAnimated therefore no longer has a separate "#ifdef
SK_HAS_WUFFS_LIBRARY" section. The first frame of the test's GIF image
data, being the required frame of the third frame, no longer has an
invalid LZW literal width according to Wuffs.
Bug: skia:8235
Change-Id: Ie94537f5232128ffc1d1547f4c0b84992e54ab02
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/226476
Commit-Queue: Leon Scroggins <scroggo@google.com>
Reviewed-by: Leon Scroggins <scroggo@google.com>
This reverts commit abb5a315af.
Two fixes:
1) GMs pass valid rowBytes when calling directly to GrGpu.
2) Check for non-null data before trying to set UNPACK_ROW_LENGTH
Bug: chromium:981254
Change-Id: I24e46b0d2b14562d6b84a29fefe3410ce5c06c94
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/226498
Reviewed-by: Greg Daniel <egdaniel@google.com>
Commit-Queue: Brian Salomon <bsalomon@google.com>
I think this is the minimum rule that's easy to understand when writing
SkSL for the interpreter that ensures we'll be able to statically
determine total stack usage of a particular function.
While writing the new test, I also noticed that we still return
(invalid) byte code, even when there are errors. Fixed that.
Change-Id: I625a8592c9ba1656074e5f0d4227d41968af7b37
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/226218
Reviewed-by: Mike Klein <mtklein@google.com>
Commit-Queue: Brian Osman <brianosman@google.com>
No longer used?
Change-Id: Ib944e5a18feb70444b34f5aba5f5c4019df1abdb
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/226217
Commit-Queue: Florin Malita <fmalita@chromium.org>
Commit-Queue: Mike Klein <mtklein@google.com>
Auto-Submit: Florin Malita <fmalita@chromium.org>
Reviewed-by: Mike Klein <mtklein@google.com>
The switch to GrColorType does mean that we can no longer represent compressed backend formats in the Mock backend surfaces.
This will require a Chrome CL before it can land in Skia.
TBR=bsalomon@google.com
Change-Id: Ie4e2d4826f960664a21d3de79933eb1cb5d06896
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/225538
Reviewed-by: Brian Salomon <bsalomon@google.com>
Reviewed-by: Greg Daniel <egdaniel@google.com>
Commit-Queue: Robert Phillips <robertphillips@google.com>
This reverts commit 0437f0f5d8.
Reason for revert: layout tests have been suppressed
Original change's description:
> Revert "More optimizations for CropToRect, for axis-aligned quads that have not been inverted"
>
> This reverts commit 88a64b4696.
>
> Reason for revert: Chrome layout tests
>
> Original change's description:
> > More optimizations for CropToRect, for axis-aligned quads that have not been inverted
> >
> > Prior to CropToRect(), GrRTC::crop_filled_rect only operated on SkRects. The
> > quad cropping code generalized the optimization to any axis-aligned quad, but
> > a consequence of this is the code had to be robust to flips and 90 degree
> > rotations. While it can handle more cases, it has lead to a 1-10% performance
> > regression on cropped-rectangle-heavy perf tests in chromium.
> >
> > This change brings back the simplest cropping solution when the axis-aligned
> > quads have not been flipped/rotated, but the general version still exists for
> > the other class of quads as well.
> >
> > Bug: chromium:980608
> > Change-Id: I83d71075cacc3d849fd9aac6436ea3244a0ae4b9
> > Reviewed-on: https://skia-review.googlesource.com/c/skia/+/225724
> > Commit-Queue: Brian Salomon <bsalomon@google.com>
> > Auto-Submit: Michael Ludwig <michaelludwig@google.com>
> > Reviewed-by: Brian Salomon <bsalomon@google.com>
>
> TBR=bsalomon@google.com,robertphillips@google.com,michaelludwig@google.com
>
> Change-Id: I428cbefa5985c6160df0460d38b0698b43d289de
> No-Presubmit: true
> No-Tree-Checks: true
> No-Try: true
> Bug: chromium:980608
> Reviewed-on: https://skia-review.googlesource.com/c/skia/+/225733
> Reviewed-by: Michael Ludwig <michaelludwig@google.com>
> Commit-Queue: Michael Ludwig <michaelludwig@google.com>
TBR=bsalomon@google.com,robertphillips@google.com,michaelludwig@google.com
Change-Id: Ibef8e521f45111b3307731966e19ef66824567e2
No-Presubmit: true
No-Tree-Checks: true
No-Try: true
Bug: chromium:980608
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/226177
Reviewed-by: Michael Ludwig <michaelludwig@google.com>
Commit-Queue: Michael Ludwig <michaelludwig@google.com>
Change-Id: I2aae762cb63fc940536959b3f0f73df472bcac7e
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/226083
Commit-Queue: Brian Salomon <bsalomon@google.com>
Commit-Queue: Greg Daniel <egdaniel@google.com>
Auto-Submit: Brian Salomon <bsalomon@google.com>
Reviewed-by: Greg Daniel <egdaniel@google.com>
This reverts commit b6b966aa43.
Change-Id: I58c34fc8520384157c5adb0056f3568b8ef1d844
Bug: skia:9230
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/225939
Reviewed-by: Mike Klein <mtklein@google.com>
Commit-Queue: Brian Osman <brianosman@google.com>
Bug: skia:
Change-Id: I2e7fc5463ea73baa82e6f1a3340e9c88d1b8be32
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/226081
Reviewed-by: Greg Daniel <egdaniel@google.com>
Commit-Queue: Jim Van Verth <jvanverth@google.com>
Also don't use GrPixelConfig to create the VkImage.
Bug: skia:7959
Bug: skia:6718
Change-Id: Ia13c5ed2fbe0542c060b725694eff9d566c491f0
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/226078
Reviewed-by: Greg Daniel <egdaniel@google.com>
Commit-Queue: Brian Salomon <bsalomon@google.com>
Bug: chromium:951893
This will help determine which piece of code left memory uninitialized.
Add a test that exercises all the different ways we might pass memory
to jpeg_write_scanlines.
Change-Id: I6392a414795da9b0471e8cd6b373a7fff8f0a1b1
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/225098
Reviewed-by: Mike Klein <mtklein@google.com>
Commit-Queue: Leon Scroggins <scroggo@google.com>
This reverts commit 5cfbee286d.
Reason for revert: Breaking DM in Google3
Original change's description:
> Make the SkSL interpreter opt-in
>
> Bug: skia:9230
>
> Change-Id: I50b707a7dd372882a983a4570810dae05c2a3638
> Reviewed-on: https://skia-review.googlesource.com/c/skia/+/225723
> Reviewed-by: Mike Klein <mtklein@google.com>
> Commit-Queue: Brian Osman <brianosman@google.com>
TBR=mtklein@google.com,brianosman@google.com
Change-Id: I5ba231ab75dd44cc1d22d497f9d7e4b732e3aa14
No-Presubmit: true
No-Tree-Checks: true
No-Try: true
Bug: skia:9230
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/225938
Reviewed-by: Brian Osman <brianosman@google.com>
Commit-Queue: Brian Osman <brianosman@google.com>
Bug: skia:9225 chromium:527565
Change-Id: I7c543bce011c247cd93cad90aca15c9754678202
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/225731
Reviewed-by: Brian Salomon <bsalomon@google.com>
Commit-Queue: Brian Osman <brianosman@google.com>
This reverts commit 88a64b4696.
Reason for revert: Chrome layout tests
Original change's description:
> More optimizations for CropToRect, for axis-aligned quads that have not been inverted
>
> Prior to CropToRect(), GrRTC::crop_filled_rect only operated on SkRects. The
> quad cropping code generalized the optimization to any axis-aligned quad, but
> a consequence of this is the code had to be robust to flips and 90 degree
> rotations. While it can handle more cases, it has lead to a 1-10% performance
> regression on cropped-rectangle-heavy perf tests in chromium.
>
> This change brings back the simplest cropping solution when the axis-aligned
> quads have not been flipped/rotated, but the general version still exists for
> the other class of quads as well.
>
> Bug: chromium:980608
> Change-Id: I83d71075cacc3d849fd9aac6436ea3244a0ae4b9
> Reviewed-on: https://skia-review.googlesource.com/c/skia/+/225724
> Commit-Queue: Brian Salomon <bsalomon@google.com>
> Auto-Submit: Michael Ludwig <michaelludwig@google.com>
> Reviewed-by: Brian Salomon <bsalomon@google.com>
TBR=bsalomon@google.com,robertphillips@google.com,michaelludwig@google.com
Change-Id: I428cbefa5985c6160df0460d38b0698b43d289de
No-Presubmit: true
No-Tree-Checks: true
No-Try: true
Bug: chromium:980608
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/225733
Reviewed-by: Michael Ludwig <michaelludwig@google.com>
Commit-Queue: Michael Ludwig <michaelludwig@google.com>
Bug: skia:9230
Change-Id: I50b707a7dd372882a983a4570810dae05c2a3638
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/225723
Reviewed-by: Mike Klein <mtklein@google.com>
Commit-Queue: Brian Osman <brianosman@google.com>
Prior to CropToRect(), GrRTC::crop_filled_rect only operated on SkRects. The
quad cropping code generalized the optimization to any axis-aligned quad, but
a consequence of this is the code had to be robust to flips and 90 degree
rotations. While it can handle more cases, it has lead to a 1-10% performance
regression on cropped-rectangle-heavy perf tests in chromium.
This change brings back the simplest cropping solution when the axis-aligned
quads have not been flipped/rotated, but the general version still exists for
the other class of quads as well.
Bug: chromium:980608
Change-Id: I83d71075cacc3d849fd9aac6436ea3244a0ae4b9
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/225724
Commit-Queue: Brian Salomon <bsalomon@google.com>
Auto-Submit: Michael Ludwig <michaelludwig@google.com>
Reviewed-by: Brian Salomon <bsalomon@google.com>
This reverts commit dcb832ac63.
Reason for revert: chrome layout tests have been suppressed.
Original change's description:
> Revert "Switch drawTexture calls to use attemptQuadOptimization"
>
> This reverts commit 9b2defc3d5.
>
> Reason for revert: Likely broke chrome layout tests
>
> Original change's description:
> > Switch drawTexture calls to use attemptQuadOptimization
> >
> > Before the quad-rendering op refactor, regular filled rectangles and
> > textured rectangles used an internal crop_filled_rect function to crop
> > to the bounding box of the clip. As the quad op refactor progressed,
> > regular filled rectangle drawing went through the new attemptQuadOptimization
> > function that supported more cropping to the exact clip when axis-aligned,
> > and could be better handle complex paints.
> >
> > Following the generalized factories added to GrTextureOp in earlier CLs,
> > it is now possible to route all textured rectangle draws through the same
> > optimizations. While there are a number of public interfaces for drawing
> > rectangles and textures on GrRTC, internally there is now just drawFilledQuad
> > (for complex paints) and drawTexturedQuad(for textures, avoids GrPaint use).
> > Both of these are very similar, and share attemptQuadOptimization(), which
> > makes all rectangle-drawing code, regardless of shader/paint/etc., handled
> > in a consistent manner.
> >
> > This doesn't entirely resolve the complexities of non-AA clipping and non-AA
> > axis-aligned rectangle drawing. That needs additional logic added to the
> > GrQuadUtils::CropToRect function, but this CL is the last structural change
> > to how the ops are created, and to GrRTC to get Ganesh to that stage.
> >
> > Change-Id: Icfbd223eef6e3da0054699334725b5084aaee58a
> > Reviewed-on: https://skia-review.googlesource.com/c/skia/+/223934
> > Commit-Queue: Michael Ludwig <michaelludwig@google.com>
> > Reviewed-by: Brian Salomon <bsalomon@google.com>
>
> TBR=bsalomon@google.com,michaelludwig@google.com
>
> # Not skipping CQ checks because original CL landed > 1 day ago.
>
> Change-Id: Ide21002c336dbc3831a2b042acfec92a509ebacf
> Reviewed-on: https://skia-review.googlesource.com/c/skia/+/224639
> Reviewed-by: Michael Ludwig <michaelludwig@google.com>
> Commit-Queue: Michael Ludwig <michaelludwig@google.com>
TBR=bsalomon@google.com,michaelludwig@google.com
# Not skipping CQ checks because original CL landed > 1 day ago.
Bug: 981879
Change-Id: I32ed19d521bd7aed93da787b3de0ca830415f42f
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/225725
Commit-Queue: Michael Ludwig <michaelludwig@google.com>
Reviewed-by: Michael Ludwig <michaelludwig@google.com>
Before this fix, skia (and thus Chrome) would fail to compile on macOS when the user (developer) had a case-sensitive file-system.
So I've replaced the incorrect includes of <metal/metal.h> by <Metal/Metal.h>
Change-Id: I6ebcc0f46608f6d840d80d18e5f5baf0744a7f16
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/225776
Reviewed-by: Jim Van Verth <jvanverth@google.com>
Commit-Queue: Mike Klein <mtklein@google.com>
All the iterations with SkStrike are moving to
bulk glyph calls. Switch over PDF.
Change-Id: I6aa0a98595d1d111254ccee333fbc162cc4eb6c8
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/225427
Commit-Queue: Herb Derby <herb@google.com>
Reviewed-by: Ben Wagner <bungeman@google.com>
Reviewed-by: Hal Canary <halcanary@google.com>
+ fix bug in SkBulkGlyphMetricsAndImages where the underlying SkAutoSTArray
was not reset.
Change-Id: I94882da30e858d012aa02f79c5fdfe4bc7d2e8aa
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/225541
Reviewed-by: Mike Klein <mtklein@google.com>
Commit-Queue: Herb Derby <herb@google.com>
This reverts commit 64860e1807.
Reason for revert: Breaking Chrome roll
Original change's description:
> Assert that lazy proxies are never borrowed or wrapped
>
> Bug: skia:
> Change-Id: Ie59550e29d5212bb0af626cd7b5bee430137289a
> Reviewed-on: https://skia-review.googlesource.com/c/skia/+/225190
> Reviewed-by: Robert Phillips <robertphillips@google.com>
> Commit-Queue: Chris Dalton <csmartdalton@google.com>
TBR=robertphillips@google.com,csmartdalton@google.com
Change-Id: If1ea208cb2f338bc3b9e38bcecb860c81a6c4fc2
No-Presubmit: true
No-Tree-Checks: true
No-Try: true
Bug: skia:
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/225542
Reviewed-by: Jim Van Verth <jvanverth@google.com>
Commit-Queue: Jim Van Verth <jvanverth@google.com>
config is completely redundant. No caller really cares what the backend
format is.
Change-Id: I93f1feb3ee61db6c21b7915bab3ee3fba5656f92
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/225194
Commit-Queue: Brian Salomon <bsalomon@google.com>
Reviewed-by: Robert Phillips <robertphillips@google.com>
Add encapsulations for buffers, and strike handling. Change all the
calls in SkFont over to the new system as an example.
Change-Id: Ib3c751d4f5bb6b04c6a92bbf907f47152f48d6b0
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/225426
Commit-Queue: Herb Derby <herb@google.com>
Reviewed-by: Mike Klein <mtklein@google.com>
Also remove dead GrCaps method (getSampleCount)
Change-Id: I6641f688472fb5ce1b2c3872dd5cdc70b32c4700
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/225422
Reviewed-by: Greg Daniel <egdaniel@google.com>
Commit-Queue: Robert Phillips <robertphillips@google.com>
Change-Id: I3680ea31bf279cc6bde58c15a3dfa6b49815cf7e
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/225138
Auto-Submit: Hal Canary <halcanary@google.com>
Commit-Queue: Michael Ludwig <michaelludwig@google.com>
Reviewed-by: Michael Ludwig <michaelludwig@google.com>
In order to effectively use the explicit backend texture allocation API Chrome needs a way to use them with surface characterizations
Change-Id: Ic61eff9f3b6b0e8280481149d7c08d37a2fe7ec0
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/222781
Reviewed-by: Brian Salomon <bsalomon@google.com>
Reviewed-by: Greg Daniel <egdaniel@google.com>
Commit-Queue: Robert Phillips <robertphillips@google.com>
This is test-only code only used by SkVMTest.cpp,
so it can live there. This cuts the dependency
of SkVM on SkStream and co.
Change-Id: I7695e527b2d16e4485f8c5f4cd39bb8300e9221d
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/225321
Reviewed-by: Herb Derby <herb@google.com>
Commit-Queue: Mike Klein <mtklein@google.com>
SkMakeSpan uses function type inference to remove boilerplate
code. The converting casts simplifies dealing with T* to const T*
uses.
Change-Id: I1851e144c4e530c275710514ce30ad75a7eb94c6
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/225192
Reviewed-by: Ben Wagner <bungeman@google.com>
Commit-Queue: Herb Derby <herb@google.com>
This necessitated some Gr*Caps refactoring
Change-Id: I8d72e9645b51bcf72d58b2c2c4fbd9ee7e83ddf1
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/225184
Reviewed-by: Brian Salomon <bsalomon@google.com>
Commit-Queue: Robert Phillips <robertphillips@google.com>
This shouldn't have any practical effect since GrRTC can't be unpremul
but removes assumptions in these functions. This code will eventually
be lifted up to GrSurfaceContext which can be unpremul.
Change-Id: If8b2fb8b19d1160bc6561f6536518563f57c4351
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/225185
Reviewed-by: Robert Phillips <robertphillips@google.com>
Commit-Queue: Brian Salomon <bsalomon@google.com>
These two guards are checking if we're building for aarch64 and thus
have F16 conversion instructions, but weren't checking if we want to use
them (if we have them _and_ we're being compiled by Clang). At head
we're trying to pass a 2-byte uint16_t to a function expecting an 8-byte
uint16x4_t, etc.
Change-Id: I21f6cd2100ec81ccdd47c4ec0575107624cd7c5a
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/225257
Reviewed-by: Herb Derby <herb@google.com>
Commit-Queue: Mike Klein <mtklein@google.com>
This will allow us to fill out all the workarounds in applyDriverCorrectnessWorkarounds
instead of having additional ones in initFormatTable.
Change-Id: I2fe2d6c831d0102e7cd8188033fee14b4e709ce4
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/225176
Reviewed-by: Brian Salomon <bsalomon@google.com>
Commit-Queue: Greg Daniel <egdaniel@google.com>
Change-Id: I74b041d3f820d423601bd06805edec001a0d980e
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/225256
Reviewed-by: Greg Daniel <egdaniel@google.com>
Commit-Queue: Greg Daniel <egdaniel@google.com>
Commit-Queue: Brian Salomon <bsalomon@google.com>
Auto-Submit: Brian Salomon <bsalomon@google.com>
Make GrGpu base class validate GrMipLevel arrays and row bytes parameters.
GrCaps states whether row bytes passed to GrGpu must be tight or not and callers
are responsible for temporary buffers if needed to make tight.
Change-Id: I2c522f7bd67c86044a36b3f70e13d7dcb38b0a6b
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/224961
Reviewed-by: Greg Daniel <egdaniel@google.com>
Commit-Queue: Brian Salomon <bsalomon@google.com>
Currently the different strikes caclulate SkPackedGlyphIDs from positions and
SkGlyphIDs. This means that there are two implementations one for SkStrike
and one for SkRemoteGlyphCache. This moves all SkPackedGlyphID calculations to
the painter thereby centralizing it.
Change-Id: If9467aca01d46aa3a2d62e7acbc97a5b6a97f13b
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/224544
Commit-Queue: Herb Derby <herb@google.com>
Reviewed-by: Ben Wagner <bungeman@google.com>
This CL starts the support for properties that are dependent on both the
format and color type. For now this adds a can upload data flag which is
currenly used in combination with the format texturable flag to determine
high level "textureability" for ganesh. The only format where these two
flags diverage from eachother currently is RGBA8 format with BGRA colortype.
Change-Id: I08b419af355d698ca4e9001949be6b7fd2e1c33d
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/224959
Commit-Queue: Greg Daniel <egdaniel@google.com>
Reviewed-by: Brian Salomon <bsalomon@google.com>
Updated caps to allow MSAA on newer Intel GPUs.
This reverts commit 41f937f042.
Bug: skia:9225
Change-Id: I15bdc4d4a754945fafcc4bc7d1afaa701d6f0426
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/224958
Auto-Submit: Brian Osman <brianosman@google.com>
Reviewed-by: Ben Wagner aka dogben <benjaminwagner@google.com>
Commit-Queue: Brian Osman <brianosman@google.com>
In SkFloatToDecimal.cpp, the |output| parameter for SkFloatToDecimal()
is named |result|. Change it to |output| and change the local variable
|output| to |output_ptr|. Fix some typos along the way.
Change-Id: Id34848f4823e69fcbd5203618ea90083fc157b6c
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/225014
Commit-Queue: Hal Canary <halcanary@google.com>
Reviewed-by: Hal Canary <halcanary@google.com>
Lays the infrastructure to use mixed samples internally, and begins
using nvpr with mixed samples on the default "gl" and "gles" configs.
In this rendition, we take the simplest approach possible re: stencil
attachments. We initially create a render target without stencil
(i.e., 0 samples). Then, any time a proxy needs a stencil buffer with
more samples than its target currently has, we create and attach a new
stencil buffer. However, we never "downgrade" a render target's
stencil attachment to one with fewer samples. So if the proxy only
needs one sample and the target has many, we leave it.
Bug: skia:
Change-Id: I8558ba799ac3dee457f349f77d4517c11413c9a9
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/224456
Commit-Queue: Chris Dalton <csmartdalton@google.com>
Reviewed-by: Brian Salomon <bsalomon@google.com>
Comments and earlier code verify that the radii are not less than 1/2.
Asserting that they are therefore greater than 1/2 fails (rarely).
Change-Id: I294eb2304c087332698dd624567a8e1d286d0144
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/225037
Reviewed-by: Jim Van Verth <jvanverth@google.com>
Commit-Queue: Jim Van Verth <jvanverth@google.com>
Bug: skia:9213
Change-Id: Ia29489b00bcaaf17d4c7b6613ed4f22e87a87ffc
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/224545
Reviewed-by: Herb Derby <herb@google.com>
Commit-Queue: Jim Van Verth <jvanverth@google.com>
Unify the handling of glyph metrics data in SkFont.
Change-Id: Ie1ad2a96ba205c0ee4570d70461b9db0cd874918
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/224577
Reviewed-by: Ben Wagner <bungeman@google.com>
Commit-Queue: Herb Derby <herb@google.com>
Change-Id: I33332967bba0f16a73633f13ffa851e38eba100d
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/224737
Reviewed-by: Brian Osman <brianosman@google.com>
Commit-Queue: Ethan Nicholas <ethannicholas@google.com>
This reverts commit e0b2dafeb6.
Reason for revert: Need to blacklist more unit tests
Original change's description:
> Test MSAA on Gen9 Intel GPUs
>
> Change-Id: Ie5441350fc150356ffb7d6acffe1089ca1de4364
> Bug: skia:9225
> Reviewed-on: https://skia-review.googlesource.com/c/skia/+/224539
> Reviewed-by: Ben Wagner aka dogben <benjaminwagner@google.com>
> Commit-Queue: Brian Osman <brianosman@google.com>
TBR=bsalomon@google.com,benjaminwagner@google.com,brianosman@google.com
Change-Id: I49c7a2f1132baeef80fa5296c2f2313f2bef0830
No-Presubmit: true
No-Tree-Checks: true
No-Try: true
Bug: skia:9225
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/224956
Reviewed-by: Brian Osman <brianosman@google.com>
Commit-Queue: Brian Osman <brianosman@google.com>
No real pressing reason to have an AVX2-specialized interpreter now that
we've got an AVX2 JIT. Keeping things centralized makes it easier to
keep track of and helps reduce dependendcies on the rest of Skia, which
in turn makes it easier to develop SkVM on wimpy machines like an RPi.
Change-Id: Ic4729603ec6c9141929b3e2a56ba380240ea5e0d
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/224822
Reviewed-by: Herb Derby <herb@google.com>
Commit-Queue: Mike Klein <mtklein@google.com>
We deduce whether to premul or unpremul based on the the input/output
alpha types. This means we also now support unpremuling on write and
premuling on read.
Class-ify former struct GrPixelInfo. Remove origin and instead pass a
flip bool to GrConvertPixels.
Unifies read/write methods on GrSurfaceContext via automatic conversion
of SkImageInfo to GrPixelInfo and making GrDirectContext an optional
parameter.
Bug: skia:7580
Change-Id: I42f6997852b4b902fb81264c6de68ca9537606aa
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/224281
Reviewed-by: Brian Osman <brianosman@google.com>
Commit-Queue: Brian Salomon <bsalomon@google.com>
Some small refactoring to common up redundant opcode building.
Oddly, I think I've got better codegen than what Clang would do here.
Clang doesn't generate uxtl-based code to unpack 8-bit to 32-bit,
instead preferring to load each byte one at a time and insert them one
at a time.
Me:
ldr s0, [x0]
uxtl v0.8h, v0.8b
uxtl v0.4s, v0.8h
Clang:
ldrb w8, [x0]
ldrb w9, [x0, #1]
ldrb w10, [x0, #2]
ldrb w11, [x0, #3]
fmov s0, w8
mov v0.s[1], w9
mov v0.s[2], w10
mov v0.s[3], w11
Change-Id: I0fdf5c6cdcde6a4eb9290936284fd3ffcb2159f6
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/224821
Reviewed-by: Herb Derby <herb@google.com>
Commit-Queue: Mike Klein <mtklein@google.com>
Although the main change in this CL is the addition of GrCaps::areColorTypeAndFormatCompatible.
This is split out of:
https://skia-review.googlesource.com/c/skia/+/222781 (Add bridge between GrContext::createBackendTexture and SkSurface::MakeFromBackendTexture)
Change-Id: I2e50fff91eb07fb1358840e1a4a76dc138a2f195
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/223932
Reviewed-by: Greg Daniel <egdaniel@google.com>
Commit-Queue: Robert Phillips <robertphillips@google.com>
This reverts commit 9b2defc3d5.
Reason for revert: Likely broke chrome layout tests
Original change's description:
> Switch drawTexture calls to use attemptQuadOptimization
>
> Before the quad-rendering op refactor, regular filled rectangles and
> textured rectangles used an internal crop_filled_rect function to crop
> to the bounding box of the clip. As the quad op refactor progressed,
> regular filled rectangle drawing went through the new attemptQuadOptimization
> function that supported more cropping to the exact clip when axis-aligned,
> and could be better handle complex paints.
>
> Following the generalized factories added to GrTextureOp in earlier CLs,
> it is now possible to route all textured rectangle draws through the same
> optimizations. While there are a number of public interfaces for drawing
> rectangles and textures on GrRTC, internally there is now just drawFilledQuad
> (for complex paints) and drawTexturedQuad(for textures, avoids GrPaint use).
> Both of these are very similar, and share attemptQuadOptimization(), which
> makes all rectangle-drawing code, regardless of shader/paint/etc., handled
> in a consistent manner.
>
> This doesn't entirely resolve the complexities of non-AA clipping and non-AA
> axis-aligned rectangle drawing. That needs additional logic added to the
> GrQuadUtils::CropToRect function, but this CL is the last structural change
> to how the ops are created, and to GrRTC to get Ganesh to that stage.
>
> Change-Id: Icfbd223eef6e3da0054699334725b5084aaee58a
> Reviewed-on: https://skia-review.googlesource.com/c/skia/+/223934
> Commit-Queue: Michael Ludwig <michaelludwig@google.com>
> Reviewed-by: Brian Salomon <bsalomon@google.com>
TBR=bsalomon@google.com,michaelludwig@google.com
# Not skipping CQ checks because original CL landed > 1 day ago.
Change-Id: Ide21002c336dbc3831a2b042acfec92a509ebacf
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/224639
Reviewed-by: Michael Ludwig <michaelludwig@google.com>
Commit-Queue: Michael Ludwig <michaelludwig@google.com>
This call is only used by the bitmap drawing. Just use SkStrike
for bitmap drawing.
Change-Id: Ib9313b21bb2f9c7de2b8ac084d905538ba45f51b
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/224576
Reviewed-by: Ben Wagner <bungeman@google.com>
Commit-Queue: Herb Derby <herb@google.com>
Before the quad-rendering op refactor, regular filled rectangles and
textured rectangles used an internal crop_filled_rect function to crop
to the bounding box of the clip. As the quad op refactor progressed,
regular filled rectangle drawing went through the new attemptQuadOptimization
function that supported more cropping to the exact clip when axis-aligned,
and could be better handle complex paints.
Following the generalized factories added to GrTextureOp in earlier CLs,
it is now possible to route all textured rectangle draws through the same
optimizations. While there are a number of public interfaces for drawing
rectangles and textures on GrRTC, internally there is now just drawFilledQuad
(for complex paints) and drawTexturedQuad(for textures, avoids GrPaint use).
Both of these are very similar, and share attemptQuadOptimization(), which
makes all rectangle-drawing code, regardless of shader/paint/etc., handled
in a consistent manner.
This doesn't entirely resolve the complexities of non-AA clipping and non-AA
axis-aligned rectangle drawing. That needs additional logic added to the
GrQuadUtils::CropToRect function, but this CL is the last structural change
to how the ops are created, and to GrRTC to get Ganesh to that stage.
Change-Id: Icfbd223eef6e3da0054699334725b5084aaee58a
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/223934
Commit-Queue: Michael Ludwig <michaelludwig@google.com>
Reviewed-by: Brian Salomon <bsalomon@google.com>
This is a temporary fix until I can figure out where these errors are
coming from.
Bug: skia:9213
Change-Id: Iac21415bf26f39f5237e12ec146769252e833d36
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/223981
Reviewed-by: Derek Sollenberger <djsollen@google.com>
Commit-Queue: Jim Van Verth <jvanverth@google.com>
Change-Id: I526cd01be501bf356307f00deb2c3067ab16de8f
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/223927
Commit-Queue: Michael Ludwig <michaelludwig@google.com>
Reviewed-by: Brian Salomon <bsalomon@google.com>
As described in the comments, this is a workaround for a rare situation
when trying to use cached program binaries in conjunction with ANGLE and
path rendering.
Change-Id: Ief5bb0042f5ec13cbea2242ff9c4f0b9f9e65641
Bug: angle:3619
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/224258
Reviewed-by: Brian Salomon <bsalomon@google.com>
Commit-Queue: Brian Osman <brianosman@google.com>
Just realized the mapping from our bytes() control nibble
to vpshufb's control byte is even easier than we thought.
Change-Id: I86fec8d3ca091ecc3fbbd59fe51ac128f0cef2fa
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/224316
Reviewed-by: Herb Derby <herb@google.com>
Commit-Queue: Mike Klein <mtklein@google.com>
This reverts commit 838c8f65ca.
Reason for revert: <INSERT REASONING HERE>
Original change's description:
> Revert "Refactor of GrGLCaps to move some code around."
>
> This reverts commit ed03e5f34b.
>
> Reason for revert: breaking win10 angle srgb
>
> Original change's description:
> > Refactor of GrGLCaps to move some code around.
> >
> > The big changes here are that in initConfigTable we were setting a few
> > member variables (srgb support and compression related) whereas all other
> > member variables were set elsewhere. So this change moves the logic out
> > of initConfigTable and places it with other similar code.
> >
> > Next setting of fSRGBWriteControl was affected by a driver workaround. We
> > need to apply this workaround before calling the initConfigTable. Thus I
> > moved the call to applyDriverWorkarounds above the initConfigTable. The
> > end result is that by the time initConfigTable is called we've now made all
> > decisions about what we will enable and disable (in terms of member variables),
> > and the only values writted in the function is setting values of the ConfigTable
> > itself.
> >
> > Change-Id: I65b1bb4e9fc49d6020e1eb58d838f47057e6d1c8
> > Reviewed-on: https://skia-review.googlesource.com/c/skia/+/224191
> > Reviewed-by: Brian Salomon <bsalomon@google.com>
> > Commit-Queue: Greg Daniel <egdaniel@google.com>
>
> TBR=egdaniel@google.com,bsalomon@google.com
>
> Change-Id: Ib9ea36b8cba28df3fdf5fcdd828315cccfcc6ce3
> No-Presubmit: true
> No-Tree-Checks: true
> No-Try: true
> Reviewed-on: https://skia-review.googlesource.com/c/skia/+/224256
> Reviewed-by: Greg Daniel <egdaniel@google.com>
> Commit-Queue: Greg Daniel <egdaniel@google.com>
TBR=egdaniel@google.com,bsalomon@google.com
Change-Id: I04d3399e021a9eed98940061f9cb4c720580dee8
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/224259
Reviewed-by: Greg Daniel <egdaniel@google.com>
Commit-Queue: Greg Daniel <egdaniel@google.com>
There are a few places that have been checking whether the *proxy*
needs stencil, in order to determine if the current render target
context needs stencil. This is problematic since a render target
context can not require stencil itself, but wrap an existing proxy
that already has stencil.
Bug: skia:
Change-Id: I2719dd3a9df15fef3d64f991cda4fadea23266bb
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/223970
Reviewed-by: Michael Ludwig <michaelludwig@google.com>
Reviewed-by: Robert Phillips <robertphillips@google.com>
Commit-Queue: Chris Dalton <csmartdalton@google.com>
This adds a new factory for GrTextureOp that generalizes the current
Make() and MakeQuad(). A following CL will remove those calls.
This also moves the texture-op emulation with GrFillRectOp into this
factory call. I found this to make more sense, both because it makes
op creation callsites simpler (drawTexturedQuad, drawTexture, and
drawTextureSet don't have to check the blend mode anymore). Additionally
the logic for matching the texture op shading behavior to a GrPaint
now lives inside GrTextureOp.
To allow for this, the to-be-removed existing factories have had the
blend mode added to their arguments as well.
Change-Id: Icda346cd203ae8caa7f5ed762ed4e0a214084dda
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/223924
Commit-Queue: Michael Ludwig <michaelludwig@google.com>
Reviewed-by: Brian Salomon <bsalomon@google.com>
- this is enough to run some of the test programs
- a couple instructions TODO: load8, store8, bytes,
so I32_SWAR and I32 are not yet JIT'd, but F32
and I32_Naive are
Change-Id: I1d97cff24aa7307726c722cce45896ed319877e7
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/224194
Reviewed-by: Herb Derby <herb@google.com>
Commit-Queue: Mike Klein <mtklein@google.com>
This reverts commit ed03e5f34b.
Reason for revert: breaking win10 angle srgb
Original change's description:
> Refactor of GrGLCaps to move some code around.
>
> The big changes here are that in initConfigTable we were setting a few
> member variables (srgb support and compression related) whereas all other
> member variables were set elsewhere. So this change moves the logic out
> of initConfigTable and places it with other similar code.
>
> Next setting of fSRGBWriteControl was affected by a driver workaround. We
> need to apply this workaround before calling the initConfigTable. Thus I
> moved the call to applyDriverWorkarounds above the initConfigTable. The
> end result is that by the time initConfigTable is called we've now made all
> decisions about what we will enable and disable (in terms of member variables),
> and the only values writted in the function is setting values of the ConfigTable
> itself.
>
> Change-Id: I65b1bb4e9fc49d6020e1eb58d838f47057e6d1c8
> Reviewed-on: https://skia-review.googlesource.com/c/skia/+/224191
> Reviewed-by: Brian Salomon <bsalomon@google.com>
> Commit-Queue: Greg Daniel <egdaniel@google.com>
TBR=egdaniel@google.com,bsalomon@google.com
Change-Id: Ib9ea36b8cba28df3fdf5fcdd828315cccfcc6ce3
No-Presubmit: true
No-Tree-Checks: true
No-Try: true
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/224256
Reviewed-by: Greg Daniel <egdaniel@google.com>
Commit-Queue: Greg Daniel <egdaniel@google.com>
Change-Id: I37eadc5800665af5ee862c4f108fa7c1f5000fa3
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/224195
Auto-Submit: Mike Klein <mtklein@google.com>
Reviewed-by: Ben Wagner <bungeman@google.com>
Commit-Queue: Ben Wagner <bungeman@google.com>
Commit-Queue: Mike Klein <mtklein@google.com>
The big changes here are that in initConfigTable we were setting a few
member variables (srgb support and compression related) whereas all other
member variables were set elsewhere. So this change moves the logic out
of initConfigTable and places it with other similar code.
Next setting of fSRGBWriteControl was affected by a driver workaround. We
need to apply this workaround before calling the initConfigTable. Thus I
moved the call to applyDriverWorkarounds above the initConfigTable. The
end result is that by the time initConfigTable is called we've now made all
decisions about what we will enable and disable (in terms of member variables),
and the only values writted in the function is setting values of the ConfigTable
itself.
Change-Id: I65b1bb4e9fc49d6020e1eb58d838f47057e6d1c8
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/224191
Reviewed-by: Brian Salomon <bsalomon@google.com>
Commit-Queue: Greg Daniel <egdaniel@google.com>
Bug: chromium:979178
Change-Id: I2cacba922c1efb0bc9fce6dfc13bd1119fc4676c
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/224193
Reviewed-by: Brian Salomon <bsalomon@google.com>
Commit-Queue: Jim Van Verth <jvanverth@google.com>
Previously, the metadata for an op entry stored the src rect and a bool.
When true, the src rect represented both the source texture coordinates
and the domain rectangle. In addition, if the real source coordinates
were actually a quadrilateral, they were stored as an optional local quad
in the buffer.
This changes it so that the metadata rect represents only the domain and
the source texture coordinates are always stored in the quad buffer.
This is slightly more storage overhead for the simple case but did not
lead to performance issues locally. It also lets us (in a later CL), fix
a bug where the src rect gets clipped in GrRTC and that clipped rect is
used as both the domain and src (when it should be used just as the src
and the original, unclipped rect is the domain).
Change-Id: Id1bcdf7c3045c330a51973a006411a9169777a91
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/223922
Commit-Queue: Michael Ludwig <michaelludwig@google.com>
Reviewed-by: Brian Salomon <bsalomon@google.com>
In the code prior to this change, if the compiler evaluated the arguments
from left to right everything would work. If the compiler evaluated the
arguments from right to left the the data was released before the calls
to get the location and size of the data. Since the compiler saw that
the release could happen before the other fields were read, it emitted
instructions like 'mov edx,dword ptr [20h]', directly inlining the
nullptr dereference.
BUG: chromium:979161
Change-Id: I0c45d9f6b8c8cd7b6c751f2b03f10aa5ea9d685c
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/224192
Reviewed-by: Hal Canary <halcanary@google.com>
Commit-Queue: Ben Wagner <bungeman@google.com>
This is a reland of c0519233cd
Original change's description:
> Reland "Separate compressed and uncompressed texture functions"
>
> This is a reland of 9acfb33ad8
>
> Original change's description:
> > Separate compressed and uncompressed texture functions
> >
> > Change-Id: Iccf31e1e4dbebde8aab4bb9b57cfb0341bb05912
> > Reviewed-on: https://skia-review.googlesource.com/c/skia/+/223802
> > Reviewed-by: Greg Daniel <egdaniel@google.com>
> > Commit-Queue: Brian Salomon <bsalomon@google.com>
>
> Change-Id: I9f212b7d34cf43216f7d2ec63b959b75fd6a71b3
> Reviewed-on: https://skia-review.googlesource.com/c/skia/+/223992
> Reviewed-by: Brian Salomon <bsalomon@google.com>
> Commit-Queue: Brian Salomon <bsalomon@google.com>
Change-Id: I0654a49dadfb56ad276051c8632b91da05bf24cd
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/224181
Commit-Queue: Brian Salomon <bsalomon@google.com>
Reviewed-by: Greg Daniel <egdaniel@google.com>
Reviewed-by: Brian Salomon <bsalomon@google.com>
This backports a little bit of the work we've done in SkVM
to the pre-compiled blit_row_s32a_opaque routine for AVX2.
Mostly I've just rewritten the function with new comments for clarity.
The key improvement is to use _mm256_shuffle_epi8 to create two copies
of the src alpha. The old version mimicked the idiom from the SSE2 code
where a byte-shuffle was not possible, instead using a shift-shift-or
pattern to construct it in three instructions. Despite being a slower
instruction, using one vpshufb does seem to beat the compound operation
by a little bit, both in practice (nanobench) and theory (llvm-mca).
This is pretty small stuff, about 0.01ns per pixel saved of ~0.15ns.
Change-Id: Ic7206e391844becab2a79a7ba6ddf4be2f82a06e
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/224185
Commit-Queue: Mike Klein <mtklein@google.com>
Commit-Queue: Florin Malita <fmalita@chromium.org>
Reviewed-by: Herb Derby <herb@google.com>
Reviewed-by: Florin Malita <fmalita@chromium.org>
Auto-Submit: Mike Klein <mtklein@google.com>
This is a reland of 076e5d3a4c
Original change's description:
> add operator "" _mask() to make Assembler clearer
>
> Change-Id: I145f0f60288eef368fc7648b25661356cb6e9f12
> Reviewed-on: https://skia-review.googlesource.com/c/skia/+/223980
> Commit-Queue: Mike Klein <mtklein@google.com>
> Reviewed-by: Herb Derby <herb@google.com>
Change-Id: I10f8301888d83ba186a32c251b25a3fbe297a6a4
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/224183
Reviewed-by: Mike Klein <mtklein@google.com>
Reviewed-by: Herb Derby <herb@google.com>
Commit-Queue: Mike Klein <mtklein@google.com>
It's a bit too easy to trip up the MSRTC bots with size_t.
(The only reason I used size_t in the first place is that
these really are byte offsets... kind of the one place we
use size_t.)
Cq-Include-Trybots: skia.primary:Test-Win2016-MSVC-GCE-CPU-AVX2-x86_64-Debug-All-MSRTC
Change-Id: I3426cec68af298a18139118b866b70b33ee34d9b
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/224182
Commit-Queue: Mike Klein <mtklein@google.com>
Commit-Queue: Brian Osman <brianosman@google.com>
Auto-Submit: Mike Klein <mtklein@google.com>
Reviewed-by: Brian Osman <brianosman@google.com>
This is a reland of 9acfb33ad8
Original change's description:
> Separate compressed and uncompressed texture functions
>
> Change-Id: Iccf31e1e4dbebde8aab4bb9b57cfb0341bb05912
> Reviewed-on: https://skia-review.googlesource.com/c/skia/+/223802
> Reviewed-by: Greg Daniel <egdaniel@google.com>
> Commit-Queue: Brian Salomon <bsalomon@google.com>
Change-Id: I9f212b7d34cf43216f7d2ec63b959b75fd6a71b3
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/223992
Reviewed-by: Brian Salomon <bsalomon@google.com>
Commit-Queue: Brian Salomon <bsalomon@google.com>
This reverts commit ff95f6ca9d.
Reason for revert: Broke command buffer GLBackendAllocationTest??
Original change's description:
> Removed made-up kSBGRA pixel config.
>
> We made up this pixel config and don't actually use it ourselves so lets
> kill it for simplicity.
>
> Change-Id: I6ae1c78fe7ada336a2411d295e8836dfeecb2d5c
> Reviewed-on: https://skia-review.googlesource.com/c/skia/+/223979
> Reviewed-by: Robert Phillips <robertphillips@google.com>
> Commit-Queue: Greg Daniel <egdaniel@google.com>
TBR=egdaniel@google.com,robertphillips@google.com,brianosman@google.com
Change-Id: I62e954495a702c7ad050719d8a1d6c4abcea3f60
No-Presubmit: true
No-Tree-Checks: true
No-Try: true
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/223990
Reviewed-by: Brian Salomon <bsalomon@google.com>
Commit-Queue: Brian Salomon <bsalomon@google.com>
Change-Id: Ic5ce56a98cfc21261ee9f5987f59db7d7c9fe38f
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/223931
Reviewed-by: Mike Klein <mtklein@google.com>
Commit-Queue: Brian Salomon <bsalomon@google.com>
We made up this pixel config and don't actually use it ourselves so lets
kill it for simplicity.
Change-Id: I6ae1c78fe7ada336a2411d295e8836dfeecb2d5c
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/223979
Reviewed-by: Robert Phillips <robertphillips@google.com>
Commit-Queue: Greg Daniel <egdaniel@google.com>
If we have the BindUniformLocation extension, we normally use that to
tell GL what location to use for each uniform, before linking. But with
cached binaries, the result is already linked, and we still need to ask
for locations.
Change-Id: Ia29f1faef9d3c9354b92cfb34332854d6c6be1b7
Bug: chromium:977938
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/223982
Reviewed-by: Brian Salomon <bsalomon@google.com>
Commit-Queue: Brian Osman <brianosman@google.com>
Change-Id: I5cc391e8d143032893511695961f5251f40e8291
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/223803
Reviewed-by: Brian Salomon <bsalomon@google.com>
Commit-Queue: Michael Ludwig <michaelludwig@google.com>
Change-Id: I4ec74a358e11d85c6eaf1658efd6f1391ee5e387
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/223930
Commit-Queue: Hal Canary <halcanary@google.com>
Commit-Queue: Ben Wagner <bungeman@google.com>
Auto-Submit: Hal Canary <halcanary@google.com>
Reviewed-by: Ben Wagner <bungeman@google.com>
All the scalers are friends, and can still access the fields, but
clients of SkGlyph can no longer access them.
Change-Id: Idbc26de74ceebeac37fa8fec9277ecf8b870e5e9
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/223801
Reviewed-by: Ben Wagner <bungeman@google.com>
Commit-Queue: Herb Derby <herb@google.com>
The majority of our gm testing has been disabling nvpr, which doesn't
match our real-world behavior where we use nvpr whenever available.
This CL fixes the issue by completely removing the explicit nvpr
configs. Now if we have nvpr, you get it.
This CL also lowers the nvpr priority in the path renderer chain and
adds a "NonNVPR" job on Quadro where we can continue to test our
non-nvpr codepaths on NVIDIA.
Bug: skia:
Change-Id: I6a36f1101c8218adcaaf10cab25d2c28e70371f0
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/223828
Reviewed-by: Brian Salomon <bsalomon@google.com>
Commit-Queue: Chris Dalton <csmartdalton@google.com>
Correctly distinguishes some cases that were previously intertwined.
Changes all the renderer tags to correspond to Intel's GPU families
(this is the same breakdown used by ANGLE, Chromium, etc.)
Extended the parsing to handle several newer families.
Change-Id: I3cd354f434aa90f79c01de4dbe98c60e86055b4b
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/223777
Reviewed-by: Brian Salomon <bsalomon@google.com>
Commit-Queue: Brian Osman <brianosman@google.com>
Confusing to get them when not working on SkVM.
Change-Id: Ib4ee35d3adcb50292f7cb5de020ddac35fb8281c
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/223918
Auto-Submit: Mike Klein <mtklein@google.com>
Reviewed-by: Mike Klein <mtklein@google.com>
Reviewed-by: Michael Ludwig <michaelludwig@google.com>
Commit-Queue: Mike Klein <mtklein@google.com>
The shared Renderer/GPU code only uses the bulk API now.
Change-Id: I04f4f75c94e599a961e10c48a72248bc2624c902
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/223800
Commit-Queue: Mike Klein <mtklein@google.com>
Reviewed-by: Mike Klein <mtklein@google.com>
This is pulled out of:
https://skia-review.googlesource.com/c/skia/+/222781 (Add bridge between GrContext::createBackendTexture and SkSurface::MakeFromBackendTexture)
Change-Id: I0c26bba2592a6e6fe759b49de83f994b05d862a2
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/223698
Reviewed-by: Brian Salomon <bsalomon@google.com>
Reviewed-by: Greg Daniel <egdaniel@google.com>
Commit-Queue: Robert Phillips <robertphillips@google.com>
This is part of bridging the explicit backend surface API and making SkSurfaces
This is pulled out of:
https://skia-review.googlesource.com/c/skia/+/222781/ (Add bridge between GrContext::createBackendTexture and SkSurface::MakeFromBackendTexture)
Change-Id: Ib55bcd8a0d1a049f230314a8f8ba7a3951b06d5c
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/223707
Reviewed-by: Greg Daniel <egdaniel@google.com>
Reviewed-by: Brian Salomon <bsalomon@google.com>
Commit-Queue: Robert Phillips <robertphillips@google.com>
Because it mutates the glyph.
Change-Id: Ic7ce320350764454d7a76335828d398f19b149d6
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/223797
Commit-Queue: Herb Derby <herb@google.com>
Commit-Queue: Mike Klein <mtklein@google.com>
Reviewed-by: Mike Klein <mtklein@google.com>
This is largely redundant with GrPixelConfig. However, we intend to
remove GrPixelConfig.
Bug: skia:7580
Change-Id: I03d92303be832711f7821f8a97d36387c9b04a9f
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/222883
Reviewed-by: Greg Daniel <egdaniel@google.com>
Commit-Queue: Brian Salomon <bsalomon@google.com>
On Mac, Metal will trigger an xpc to a process to compile shaders and
link pipeline programs. Sometimes that process can crash or hang. By using
an asynchronous call with a timeout we can at least recover in this case.
Bug: chromium:974219
Change-Id: I179daa86979b1217458e7be210fccd5edcbffdd0
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/222884
Reviewed-by: Christopher Cameron <ccameron@chromium.org>
Reviewed-by: Greg Daniel <egdaniel@google.com>
Commit-Queue: Jim Van Verth <jvanverth@google.com>
So far this is just as easy as I had hoped.
Change-Id: I5f69a900b32d9bf70156b55e334233d7376b820f
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/223340
Reviewed-by: Herb Derby <herb@google.com>
Commit-Queue: Mike Klein <mtklein@google.com>
This call maintains image invariants while merging in
an image. It replaces much dubious code.
Change-Id: Id57fa1169f8944b2d10063fde983cd9d8256b598
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/223303
Reviewed-by: Mike Klein <mtklein@google.com>
Commit-Queue: Herb Derby <herb@google.com>
Instead of allocating into a std::vector, we do one quick first pass to
measure how much memory we need to allocate, mmap enough pages for that,
then another real writing pass.
This cuts a microsecond or so off the profile. There's another
microsecond left to cut if we could eliminate that first measuring pass,
but I'm no longer sure it's easy to come up with a good upper limit on
the program size now that I'm thinking about the data part of the
program as well.
vpshufb is our current max instruction at 9 bytes of code, but that also
implies another 32 bytes of control data. I'm not sure I feel very
clever allocating 41 * |instructions| bytes to be conservatively safe...
it seems like ridiculous overkill.
Ultimately I found it easier to just measure twice, cut once.
Change-Id: I16ccdafbc789711837b41b3d5a557808798eb1b4
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/223305
Commit-Queue: Mike Klein <mtklein@google.com>
Reviewed-by: Herb Derby <herb@google.com>
Use float* to match the ByteCode run API (and make the sizing of data
clearer). Add a lane index to all external value calls. My upcoming
overhaul of the particle code needs this, but I wanted to break that
(large) CL up.
Change-Id: I0588cd7769a1dced9f088de5756947bb744c146b
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/223178
Commit-Queue: Brian Osman <brianosman@google.com>
Reviewed-by: Mike Klein <mtklein@google.com>
Make existing trace events in GrGpu go direct to TRACE_EVENT0.
Bug: chromium:919885
Change-Id: Ica270125eb614a31860fd3ad4b78178e7bd24296
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/223336
Auto-Submit: Brian Salomon <bsalomon@google.com>
Reviewed-by: Brian Osman <brianosman@google.com>
Commit-Queue: Brian Salomon <bsalomon@google.com>
This is a reland of 731454a085
Original change's description:
> Prefer using GrOvalOpFactory over GrFillRRect for circles and
> axis-aligned circular roundrects.
>
> Bug: chromium:971936
> Change-Id: I4cd0cd9047b9b06d657826820ba5a937547f87c3
> Reviewed-on: https://skia-review.googlesource.com/c/skia/+/221000
> Commit-Queue: Jim Van Verth <jvanverth@google.com>
> Reviewed-by: Khushal Sagar <khushalsagar@chromium.org>
Bug: chromium:971936
Change-Id: I8a61cff3e065177a5b2320072b45c1a619970ff6
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/222794
Reviewed-by: Jim Van Verth <jvanverth@google.com>
Commit-Queue: Jim Van Verth <jvanverth@google.com>
Change-Id: I5428be31367fc85fb1fe4f98f4e7536f17c93f10
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/223301
Commit-Queue: Brian Salomon <bsalomon@google.com>
Commit-Queue: Brian Osman <brianosman@google.com>
Auto-Submit: Brian Salomon <bsalomon@google.com>
Reviewed-by: Brian Osman <brianosman@google.com>
Converting to glyph() style calls that return SkGlyph*. This is mainly preparation
for removing converting findImage(const SkGlyph&) to prepareImage(SkGlyph*).
+ Misc cleanups mainly fWidth -> width() type things.
Change-Id: Id5c9b0ba5856b3ea54353ece4d05fa495cc5a640
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/223187
Commit-Queue: Herb Derby <herb@google.com>
Reviewed-by: Mike Klein <mtklein@google.com>
By itself this CL isn't all that compelling but I believe we need some intermediate path to wean ourselves off of GrPixelConfig. In particular, I believe isFormatTexturable will not need an SkColorType parameter in the future.
This is pulled out of:
https://skia-review.googlesource.com/c/skia/+/222781/ (Add bridge between GrContext::createBackendTexture and SkSurface::MakeFromBackendTexture)
which adds SkSurface::isCompatible - so the SkSurface_Gpu::isCompatible calls have been removed from this CL.
Change-Id: I6c2b8a2c4af98c1437b92c58513f34014e551b2e
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/223188
Reviewed-by: Greg Daniel <egdaniel@google.com>
Commit-Queue: Robert Phillips <robertphillips@google.com>
This means GrSurfaceContext's know their alpha type.
All GrRenderTargetSurfaceContexts are kPremul.
Make GrTextureProducer store GrColorSpaceInfo.
Bug: skia:7580
Change-Id: I5ff321ef52c0edd32e5fac99dff95d44aa66f592
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/223184
Commit-Queue: Brian Salomon <bsalomon@google.com>
Reviewed-by: Greg Daniel <egdaniel@google.com>
I was just reading the ARM docs and realized that their BIC ("BIt
Clear") is the same as SSE's ANDN ("AND Not") instruction. It's kind of
a neat little tool to have laying around... comes up more than you'd
think, and it's sometimes the clearest way to express what you're doing,
as in the changed program here where the comment is "mask away the low
bits". That's a bit_clear with a mask for what you want to clear away!
And the real reason to write this up is that I want to have a CL to
point to that shows how to add an instruction top to bottom.
Change-Id: I99690ed9c1009427b3986955e7ae6264de4d215c
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/223120
Commit-Queue: Mike Klein <mtklein@google.com>
Reviewed-by: Herb Derby <herb@google.com>
Reviewed-by: Mike Reed <reed@google.com>
This CL allows user to indicate that they have a protected content in
GrVkBackendContext creation which results in protected CommandPool and Queue
usage.
Bug: skia:9016
Change-Id: I6a478d688b6988c2c5e5e98f18f58fb21f9d26ae
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/210067
Commit-Queue: Greg Daniel <egdaniel@google.com>
Auto-Submit: Emircan Uysaler <emircan@google.com>
Reviewed-by: Greg Daniel <egdaniel@google.com>
Instruction is the fundamental data, and Analysis derived from it.
The fields in Analysis are only* needed in Builder::done(), and this
split seems to help clarify what done() can tweak (Analysis) and what
it cannot (fProgram, Instructions). done() is now const.
No speed change as far as I can tell.
* As you may notice looking at the test expectations, making analysis
ephemeral means that dump() can no longer print the skull for dead code
or the arrow for hoisted. The register program that's also in the
expectation file still reflects both of these optimizations, so we're
not really losing any information. Just maybe less demo-friendly.
Change-Id: I79feb57558525591baf3faadeb59c418c12793f3
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/223119
Commit-Queue: Mike Klein <mtklein@google.com>
Reviewed-by: Herb Derby <herb@google.com>
Each one of these Instructions has its own register assignment,
so instead of allocating them in a little temporary side vector,
allocate them along with the main Program entries, just like
the other metadata, hoist and life.
No noticeable change in perf.
Change-Id: I3db8c1520d52f5787111b227e6becfef49e5a892
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/223118
Commit-Queue: Mike Klein <mtklein@google.com>
Reviewed-by: Herb Derby <herb@google.com>
This cuts the overhead bench from about 19µs to about 15µs.
The key insight here is that the only registers that might become
available after any given instruction are the ones that hold that
instruction's inputs. We can check when they become available
directly from the original Builder::Program, without needing a
side death schedule data structure.
Marking hoisted instructions as having life == program size
helps make this logic a little simpler to reason through.
Change-Id: Ifb9957f2d0e323e0e5d07996a2cc988f7c8b4c3f
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/223117
Commit-Queue: Mike Klein <mtklein@google.com>
Reviewed-by: Herb Derby <herb@google.com>
This splits the ID namespace into Reg and Val types, hopefully making it
a little easier to follow what's going on, and if we want, allowing us
to size them differently (e.g. val at i32 or i16, reg at i16 or u8). I
didn't notice any speed change when shrinking either, so I've left them
both at i32 for maximum flexibility.
I played with making these strong typedefs with both structs and enum
classes, but both felt a little awkward. I'm still open to the idea.
Change-Id: Ie0adf6944ed6254eb21dfdfb59894c4e30476443
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/223077
Commit-Queue: Mike Klein <mtklein@google.com>
Reviewed-by: Herb Derby <herb@google.com>
This cuts a field from Builder::Instruction, and also makes the code
easier to follow, I think. Now d, x, and y are always registers, and
only the final field may be a register z or an immediate.
Change-Id: I33bbe0c6fb8cb96b85f0b0e8c30df3fa4d233c1b
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/223076
Commit-Queue: Mike Klein <mtklein@google.com>
Reviewed-by: Herb Derby <herb@google.com>
From now on, sample counts always refer to the number of actual color
samples, and render targets don't have separate color and stencil
sample counts.
If mixed samples support is available when making a
"GrAAType::kCoverage" draw, then an op may attach and use a mixed
sampled stencil buffer internally. But this will all be invisible to
the client.
After this CL, we temporarily won't have a mode to use nvpr with mixed
samples. That will soon be fixed by a follow-on CL that enables nvpr
with mixed samples in the normal "gl" and "gles" configs.
Bug: skia:
Change-Id: I1cb8277f0d2d0d371f24bb9f39cd473ed5c5c83b
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/221878
Commit-Queue: Chris Dalton <csmartdalton@google.com>
Reviewed-by: Brian Salomon <bsalomon@google.com>
This new bench lets us measure the overhead of program building,
optimization, and JITting. Surprisingly, at head the optimization in
Builder::done() takes longer than the JIT.
The new bench clocks in around 40µs on my laptop at head,
then 32µs after switching val_to_reg to be an std::vector,
then 27µs after switching deaths to be an std::vector too,
then 22µs after switching fIndex to be an SkTHashMap,
then 20µs after calling program.reserve(fProgram.size()),
then 19µs after switching JIT data maps to SkTHashMap too.
I tried swapping some std::vector for SkTDArray to no benefit, actually
a little detriment. So I think this is roughly all the low-hanging
fruit, with time split now roughly equally between Builder::Done(),
JITting in Program::eval(), and the original calls to Builder
themselves.
Also disable perf dumps on Mac. No real value there until I can dump a
dylib, and it's just one more thing I have to remember to disable before
running this sort of benchmark.
Change-Id: I1c6e58ed00ac94ad622c7d740712634f60787102
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/222984
Reviewed-by: Herb Derby <herb@google.com>
Commit-Queue: Mike Klein <mtklein@google.com>
For now, disable the vpmovusdb AVX-512 instruction, using the compound
AVX2 fallback instead. I need to learn how to encode EVEX prefixes
before we can use that, and it's not very important.
That's everything! We're fully in control now, and should be able to
run this on any x86-64 Linux or Mac. And we can relax some of the
defined(SKVM_JIT) guards so that, e.g., we can unit test Assembler even
on all platforms.
Stifle some warnings about ~bool by ~(int)bool.
Would like to enable when is_mac too but can't seem to get past
(bogus?) thread annotation on the bots. My local Mac is fine. :/
Change-Id: If00bdd97ebd9684ed109933e2fa70c5e6f6ea339
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/222631
Commit-Queue: Mike Klein <mtklein@google.com>
Reviewed-by: Herb Derby <herb@google.com>
Most image filters were fixed with just the changes to SkImage::makeWithFilter
and the changes to SkSpecialImage's subset handling (particularly the
raster backend that could read from a bitmap view, or ganesh impls that relied
SkSpecialImage::draw).
The gpu implementation for alpha threshold, blurs, matrix convolutions,
and displacement maps have been updated to account for the special image's
offset when it reads from the backing texture.
Change-Id: I8778aa373e60e9268961305057b2bf6da2bdb3af
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/221121
Reviewed-by: Robert Phillips <robertphillips@google.com>
Commit-Queue: Michael Ludwig <michaelludwig@google.com>
Move the invariants for glyph image data into SkGlyph.
Change-Id: I1958612bb73cfffe42df19a11c8899048559013b
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/222876
Commit-Queue: Herb Derby <herb@google.com>
Reviewed-by: Mike Klein <mtklein@google.com>
Bug: chromium:977315
Change-Id: Ia5b734f5c0f0806af0f096de5add880a777c5c25
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/222793
Reviewed-by: Brian Salomon <bsalomon@google.com>
Commit-Queue: Michael Ludwig <michaelludwig@google.com>
This shows off a little how easy backwards-only labels are.
The rip == rbp + Mod::Indirect convention isn't something
you'd be able to guess without just looking at the docs.
I'm not actually sure if you can only use rbp or also r13,
but LLVM seems to always do the equivalent of rbp... might
just be that high bit in VEX is ignored: they're registers
5 and 13, 8 apart, only distinguished by that bit.
Convenienly RIP addressing is always 32-bit, so there's
no benefit to spending time checking whether the offset
fits in a byte, though most of our offsets would.
Change-Id: I01b7fb1500667e1bf98490d5144459f92e1b375d
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/222857
Commit-Queue: Mike Klein <mtklein@google.com>
Reviewed-by: Herb Derby <herb@google.com>
I think this makes the relatioship between mask and entry clearer?
Can't have JIT code handle >0 elements unless that JIT code itself
exists.
Change-Id: I238d54a5084c7f90bd32c83db5423840cf415b17
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/222856
Commit-Queue: Mike Klein <mtklein@google.com>
Reviewed-by: Herb Derby <herb@google.com>
This moves the responsibility for allocating executable code out of
Assembler. The pages Xbyak uses are obviously executable, so this is
redundant right now, but it'll let us switch to something simple like
std::vector<uint8_t> as we continue to cut out Xbyak.
Make how Program holds its cached JIT program slightly less of a mess.
Change-Id: I38d6f01006da1da60f4aed675e9ddf97de9aec52
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/222575
Auto-Submit: Mike Klein <mtklein@google.com>
Reviewed-by: Herb Derby <herb@google.com>
Commit-Queue: Mike Klein <mtklein@google.com>
Change-Id: I4d1a102264d8c97bf9120c3891d569ef96a92922
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/222782
Commit-Queue: Brian Osman <brianosman@google.com>
Reviewed-by: Mike Klein <mtklein@google.com>
By putting data first in descending alignment then code, we never need
any alignment padding.
This also makes all jumps and ip-relative data loads backward, so
they're really easy to assemble. No need for any sort of deferred
where-does-this-label-mean logic; the label can just be a simple byte
offset established before you need to use it.
Nothing new switched off of Xbyak in this CL, but the rearrangement
makes the rest a lot easier.
The one downside I've found so far is that the disassembly of the
first instruction can get confused into data or other instructions,
e.g.
63: 01 ff add %edi,%edi
65: 00 ff add %bh,%bh
67: 00 00 add %al,(%rax)
69: ff 00 incl (%rax)
6b: ff c4 inc %esp
6d: e2 7d loop ec <skvm-jit-884702985+0xac>
6f: 18 05 eb ff ff ff sbb %al,-0x15(%rip) # 60 <skvm-jit-884702985+0x20>
75: c4 e2 7d 18 0d e6 ff ff ff vbroadcastss -0x1a(%rip),%ymm1 # 64 <skvm-jit-884702985+0x24>
7e: c4 e2 7d 18 15 e1 ff ff ff vbroadcastss -0x1f(%rip),%ymm2 # 68 <skvm-jit-884702985+0x28>
There are 3 vbroadcastss instructions here, each starting with c4 e2 7d
18, but the first has been disassembled as if its c4 were part of the
last data entry (0xff00ff00) as inc %esp.
Probably not a big deal for now, particularly since those vbroadcastss
are all outside the loop and never show up on a profile. If it gets too
confusing I think we can dump the programs starting from the beginning
of the code instead of from the data; we won't be able to inspect the
data, but everything should disassemble perfectly.
Change-Id: I0cc864359fd0740fc026070eaf2b6cb130783a57
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/222574
Auto-Submit: Mike Klein <mtklein@google.com>
Reviewed-by: Herb Derby <herb@google.com>
Commit-Queue: Mike Klein <mtklein@google.com>
This centralizes the initial-lane-mask logic, and makes the return value
copying much more straightforward by just passing in the width. Lets us
shrink the arrays in the interpreter pipeline stage to the correct size.
Also normalize some formatting and structure.
Change-Id: I446598dcdd550d88ff1db1afe7507f31fa96d1d7
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/222510
Commit-Queue: Brian Osman <brianosman@google.com>
Reviewed-by: Mike Klein <mtklein@google.com>
Introduce textBlobToGlyphRunListWithoutRSX to convert text blob into
glyph runs. Convert the core of the code from working over text blobs
to working over glyph runs.
+ Misc cleanups
Change-Id: I33c1fc5e948dd7270031496325a96409f2cfeeb6
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/222277
Reviewed-by: Ben Wagner <bungeman@google.com>
Commit-Queue: Herb Derby <herb@google.com>
Change-Id: I1d133259264adfdc872b0f4aeaa9390363c46341
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/222040
Reviewed-by: Mike Klein <mtklein@google.com>
Commit-Queue: Brian Osman <brianosman@google.com>
Since explicit resource allocation has stuck these instantiate calls are no longer required.
Change-Id: I5a8a7fa714eb1e9550f4f645ce8fced2d5f7aa4e
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/222457
Reviewed-by: Brian Salomon <bsalomon@google.com>
Commit-Queue: Robert Phillips <robertphillips@google.com>
Like the Posix code below it, the Mac semaphore wait code
can wake spuriously. Keep trying if we get KERN_ABORTED.
Pattern aped from V8. I noticed they like us don't do
anything to test if the Windows WaitForSingleObject()
call fails. Can it fail?
Bug: chromium:977341
Change-Id: I34f407fc4d6717deb6edcf7aa7bed1f8fb8b1baa
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/222583
Reviewed-by: Herb Derby <herb@google.com>
Commit-Queue: Mike Klein <mtklein@google.com>
This wraps up everything that doesn't touch memory or a label.
Small cleanups and refactoring as I start to understand some
things better...
Change-Id: I788fa877cfcab8f87c961df28fe561b51a5c62ff
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/222571
Commit-Queue: Mike Klein <mtklein@google.com>
Reviewed-by: Herb Derby <herb@google.com>
The encoding kind of all goes through the same paths,
as the three argument instructions, but like the nursery
rhyme when there are only two they kind of all roll over
and the op-extension hops into the bed.
vpermq is the first place we need to set the W bit
to indicate a 64-bit lane operation, so a little
minimal plumbing for that. It takes its arguments
a little differently too, passing dst where you'd
expect, the source where we'd pass y, and requiring
us to pass literal 0000 for the vvvv bits in VEX
(inverted as normal to literal 1111).
Change-Id: I91a4cd1b316eb908992631ce8b2cb3c62078e8c6
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/222565
Commit-Queue: Mike Klein <mtklein@google.com>
Reviewed-by: Herb Derby <herb@google.com>
Somewhat arbitrarily switched from vandps et al. float bitwise
ops to vpand et al. int bitwise ops. We tend to use them more
int-y generally so the disassembly maybe reads more clearly?
They're identical in behavior AFAIK.
Shuffle tmp around from being an Xbyak::Ymm to its index.
I don't think there's anything tricky here. Spookily things
all seem to work first try, as long as I don't make a typo.
Change-Id: I2b5d4ded7800915824cbd7c917edfd36e229d306
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/222528
Auto-Submit: Mike Klein <mtklein@google.com>
Reviewed-by: Herb Derby <herb@google.com>
Commit-Queue: Mike Klein <mtklein@google.com>
This reverts commit 6f24faa623.
Reason for revert: larger impact on GMs than expected
Original change's description:
> GPU: always use TopLeft origin for saveLayer render targets.
>
> Should have no user-visible result.
>
> Change-Id: Iae444888557347bfdf75e6353966cde907ee0a1e
> Reviewed-on: https://skia-review.googlesource.com/c/skia/+/222506
> Reviewed-by: Greg Daniel <egdaniel@google.com>
> Commit-Queue: Stephen White <senorblanco@chromium.org>
TBR=egdaniel@google.com,bsalomon@google.com,senorblanco@chromium.org
Change-Id: Ic8217206adfbd7d506a70c3d9ca828f0dde58c66
No-Presubmit: true
No-Tree-Checks: true
No-Try: true
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/222509
Reviewed-by: Stephen White <senorblanco@chromium.org>
Commit-Queue: Stephen White <senorblanco@chromium.org>
The bufferAllocator was being unref'ed before the SkSpinlock went out
of scope, so the spinlock was then using deleted memory.
Bug: skia:8243
Change-Id: I69f090acccaaa3ba7fe2e4190103019a6f4e9359
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/222503
Commit-Queue: Jim Van Verth <jvanverth@google.com>
Reviewed-by: Robert Phillips <robertphillips@google.com>
Change-Id: I6b709a2b5e8014cedb152e763ea4470c6f79de4b
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/222504
Commit-Queue: Brian Osman <brianosman@google.com>
Commit-Queue: Mike Klein <mtklein@google.com>
Auto-Submit: Brian Osman <brianosman@google.com>
Reviewed-by: Mike Klein <mtklein@google.com>
Should have no user-visible result.
Change-Id: Iae444888557347bfdf75e6353966cde907ee0a1e
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/222506
Reviewed-by: Greg Daniel <egdaniel@google.com>
Commit-Queue: Stephen White <senorblanco@chromium.org>
Change-Id: Ia2e21bc984c509cd2405499f93ee5e19941cc492
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/222499
Commit-Queue: Brian Osman <brianosman@google.com>
Reviewed-by: Mike Klein <mtklein@google.com>
This reverts commit 731454a085.
Reason for revert: Seeing a lot of performance regressions.
Original change's description:
> Prefer using GrOvalOpFactory over GrFillRRect for circles and
> axis-aligned circular roundrects.
>
> Bug: chromium:971936
> Change-Id: I4cd0cd9047b9b06d657826820ba5a937547f87c3
> Reviewed-on: https://skia-review.googlesource.com/c/skia/+/221000
> Commit-Queue: Jim Van Verth <jvanverth@google.com>
> Reviewed-by: Khushal Sagar <khushalsagar@chromium.org>
TBR=jvanverth@google.com,csmartdalton@google.com,khushalsagar@chromium.org
# Not skipping CQ checks because original CL landed > 1 day ago.
Bug: chromium:971936
Change-Id: Iab803b1777ef5e3d754d8ac1f404b01f57d9c1a8
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/222501
Reviewed-by: Jim Van Verth <jvanverth@google.com>
Commit-Queue: Jim Van Verth <jvanverth@google.com>
Just landing the initial struct before adding specific flags to the struct
and filling out the formats.
Bug: skia:6718
Change-Id: I1013845cb61482184915d181a7fc7f85800b72c5
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/222498
Reviewed-by: Robert Phillips <robertphillips@google.com>
Commit-Queue: Greg Daniel <egdaniel@google.com>
- 32x8 i32 add,sub,mul
- add I32_Naive bench/test builder to get better i32 mul coverage
- minor refactoring all over
Change-Id: I13cc19ff37a2da0bcff289ba51baac08f456d6c5
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/222485
Reviewed-by: Herb Derby <herb@google.com>
Commit-Queue: Mike Klein <mtklein@google.com>
This extends the runStriped to all for return data.
GPU impl not done yet, will be done in a follow-on CL.
Change-Id: Ib107d2945f6fdb34ce1b5405a6c88a5ae7e9f7ac
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/221539
Reviewed-by: Brian Osman <brianosman@google.com>
Commit-Queue: Mike Reed <reed@google.com>
Buffer creation is a big hit on Metal. This allows us to recycle more
of the buffer.
Bug: skia:8243
Change-Id: If2e3b13d8175dfe370bd61771af0a59cce277340
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/221120
Commit-Queue: Jim Van Verth <jvanverth@google.com>
Reviewed-by: Brian Osman <brianosman@google.com>
This reverts commit 27239e456a.
Revert "Revert "Add function to GrDataUtils to handle color conversions.""
This reverts commit c34d993b62.
Change-Id: Iac1bdaa6f8380e63bbb87394e7fca96808572131
Bug: skia:8962
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/222039
Reviewed-by: Brian Salomon <bsalomon@google.com>
Commit-Queue: Brian Salomon <bsalomon@google.com>
I'm getting tired of having changes to SkVM.h trigger
383 build targets, now 13. No other reason.
Change-Id: I12b637e68db409ede9c0c5be74fcca66cca32505
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/222479
Commit-Queue: Mike Klein <mtklein@google.com>
Commit-Queue: Brian Osman <brianosman@google.com>
Auto-Submit: Mike Klein <mtklein@google.com>
Reviewed-by: Brian Osman <brianosman@google.com>
This CL:
enables the Y416 row (i.e., uses RGBA_16161616)
spruces up the GL caps configuration of R16, RG16, RGBA16 and RG16F
Change-Id: I405c5acaf3f31b7ab0e9797e6f5df1d4c41d70d0
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/220535
Commit-Queue: Robert Phillips <robertphillips@google.com>
Reviewed-by: Greg Daniel <egdaniel@google.com>
This CL:
replaces GrProxyRef with sk_sp
streamlines GrIORefProxy to be more like SkRefCntBase (i.e., move the fTarget pointer to GrSurfaceProxy)
Change-Id: I17d515100bb2d9104eed64269bd3bf75c1ebbbb8
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/221997
Reviewed-by: Brian Salomon <bsalomon@google.com>
Commit-Queue: Robert Phillips <robertphillips@google.com>
Before the the consolidation to the internal drawFilledQuad(),
regular draws would use the logical rt bounds to decide if it can be
a fullscreen clear. Stencil rect draws would use the worst case bounds
(needed to pass GrDefaultPathRendererTest).
Initially, the new drawFilledQuad() switched to always using the worst
case bounds to determine fullscreen clears. Since stencil rect draws
used drawFilledQuad(), this was necessary for the path renderer tests to
pass. But this had the side effect of making render targets with approx.
backing storage no longer perform fullscreen clears when a draw would
otherwise cover the logical bounds of the render target.
Normally, this would only be a performance issue, but it also seems to
have exposed a driver bug in the Nexus 5x on Vulkan where the draws
didn't properly fill the backing store. Having attemptQuadOptimization()
choose its rt bounds based on the presence of stencil settings keeps
both tests happy and increases the likelihood of using fullscreen clears.
Change-Id: I6df2c789211f32a4c94bee394d27b7fb4f7293e2
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/222278
Commit-Queue: Michael Ludwig <michaelludwig@google.com>
Commit-Queue: Greg Daniel <egdaniel@google.com>
Auto-Submit: Michael Ludwig <michaelludwig@google.com>
Reviewed-by: Greg Daniel <egdaniel@google.com>
This reverts commit 1fb0874af5.
Reason for revert: maybe breaking chrome roll.
Original change's description:
> Remove gpu config fallback in onMakeColorTypeAndColorSpace.
>
> If the client is asking for a specific colorType for a GPU image it seems
> wrong to give them one they didn't ask for.
>
> Change-Id: I65bcdfcb0f744458e2162fe266c08bd3c8fb5a85
> Reviewed-on: https://skia-review.googlesource.com/c/skia/+/222176
> Reviewed-by: Brian Osman <brianosman@google.com>
> Commit-Queue: Greg Daniel <egdaniel@google.com>
TBR=egdaniel@google.com,bsalomon@google.com,brianosman@google.com
Change-Id: If3469e45d03471af126605072f4d6debec7101b6
No-Presubmit: true
No-Tree-Checks: true
No-Try: true
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/222038
Reviewed-by: Greg Daniel <egdaniel@google.com>
Commit-Queue: Greg Daniel <egdaniel@google.com>
If the client is asking for a specific colorType for a GPU image it seems
wrong to give them one they didn't ask for.
Change-Id: I65bcdfcb0f744458e2162fe266c08bd3c8fb5a85
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/222176
Reviewed-by: Brian Osman <brianosman@google.com>
Commit-Queue: Greg Daniel <egdaniel@google.com>
This allows backends to use an expression rather than a variable for
sampler access.
Change-Id: I2346ac418c8d6277416a4430e2eaf6ab3ffee0a2
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/222036
Reviewed-by: Brian Salomon <bsalomon@google.com>
Commit-Queue: Stephen White <senorblanco@chromium.org>
This reverts commit 7694b90eb0.
Reason for revert: suppression: https://chromium-review.googlesource.com/c/chromium/src/+/1666472
Original change's description:
> Revert "Consolidate quad optimizations into single internal function."
>
> This reverts commit 646616a78f.
>
> Reason for revert: Suspected as cause of layout test changes.
>
> Original change's description:
> > Consolidate quad optimizations into single internal function.
> >
> > Routes all non-textured quad draws through single internal function
> >
> > Change-Id: Ief66864a0ad2d598982c5bf500c8a84ecbf84387
> > Reviewed-on: https://skia-review.googlesource.com/c/skia/+/215455
> > Commit-Queue: Michael Ludwig <michaelludwig@google.com>
> > Reviewed-by: Robert Phillips <robertphillips@google.com>
>
> TBR=robertphillips@google.com,michaelludwig@google.com
>
> Change-Id: I0dc6a0d948c0f5e9221ff6c9fbbbbbb9bc3d9bc0
> No-Presubmit: true
> No-Tree-Checks: true
> No-Try: true
> Reviewed-on: https://skia-review.googlesource.com/c/skia/+/221737
> Reviewed-by: Brian Salomon <bsalomon@google.com>
> Commit-Queue: Brian Salomon <bsalomon@google.com>
TBR=bsalomon@google.com,robertphillips@google.com,michaelludwig@google.com
Change-Id: I4e5d39d603d32b18c48db291fb1650fe33e9ba11
No-Presubmit: true
No-Tree-Checks: true
No-Try: true
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/222096
Commit-Queue: Michael Ludwig <michaelludwig@google.com>
Reviewed-by: Michael Ludwig <michaelludwig@google.com>
I'm hoping to have a few pinch points that will actually hold a mutex.
The call getAdvances uses to get glyphs, called metrics(), is one of them.
+ Update some call sites to use the new glyph() call.
Change-Id: I04158e200db43758fe6d33337e5d1137f197a5b7
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/221777
Commit-Queue: Herb Derby <herb@google.com>
Reviewed-by: Mike Klein <mtklein@google.com>
This reverts commit 59b22f9297.
Reason for revert: ios
Original change's description:
> Reland "Reland "Interpreter: Support striped inputs for less overhead""
>
> This reverts commit 2c59b4e9ea.
>
> Change-Id: I2b06936994430722b8fc3890ff9b4a6f4710db04
> Reviewed-on: https://skia-review.googlesource.com/c/skia/+/221998
> Reviewed-by: Brian Osman <brianosman@google.com>
> Reviewed-by: Mike Klein <mtklein@google.com>
> Commit-Queue: Brian Osman <brianosman@google.com>
TBR=mtklein@google.com,brianosman@google.com,reed@google.com
Change-Id: I321d257e400f5df3c6115dfe9aa18a2794a95910
No-Presubmit: true
No-Tree-Checks: true
No-Try: true
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/222099
Reviewed-by: Brian Osman <brianosman@google.com>
Commit-Queue: Brian Osman <brianosman@google.com>
When starting a nested loop, we can't rely on the condition mask to keep
dead lanes dead, because of conditional breaks/continues that may have
happened earlier. So always narrow the loop mask by inheriting the
previous one.
Change-Id: I5bb076e6467fe1b6a2f682a590e8ae972a440b03
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/222098
Reviewed-by: Ethan Nicholas <ethannicholas@google.com>
Reviewed-by: Mike Klein <mtklein@google.com>
Commit-Queue: Brian Osman <brianosman@google.com>
This is a reland of a224fc1105
Changes since original:
- switch fJIT_K to less error prone fJITMask
- guard fJIT Assembler Program member with SKVM_JIT
Not really sure why the mips64el-Debug bot's compiler is crashing;
it does at least make sense to crash where it does... the file
includes SkOpts.h which includes SkVM.h.
If no reasonable code transformation can get it working again
I'll remove the bot. The -Release version is fine, and mips64el
is one of those things I'd happily flush if it blocks progress.
In this end I think all this SKVM_JIT and Xbyak stuff should
go away and make things simple again, hopefully too simple to
crash GCC. :|
Original change's description:
> extract Assembler so it can be tested
>
> And start documenting some structs we'll need
> to replace xbyak.
>
> Change-Id: I21c91642799a54e10af85afc8edbe12a9b4aa062
> Reviewed-on: https://skia-review.googlesource.com/c/skia/+/221644
> Reviewed-by: Herb Derby <herb@google.com>
> Commit-Queue: Mike Klein <mtklein@google.com>
Cq-Include-Trybots: skia.primary:Test-Debian9-Clang-GCE-CPU-AVX2-x86_64-Release-All-SK_CPU_LIMIT_SSE2,Build-Debian9-GCC-mips64el-Debug
Change-Id: I6d7c27bc758b23c164ee67067cdfacc291e289fc
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/221983
Reviewed-by: Mike Klein <mtklein@google.com>
Commit-Queue: Mike Klein <mtklein@google.com>
This reverts commit 2c59b4e9ea.
Change-Id: I2b06936994430722b8fc3890ff9b4a6f4710db04
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/221998
Reviewed-by: Brian Osman <brianosman@google.com>
Reviewed-by: Mike Klein <mtklein@google.com>
Commit-Queue: Brian Osman <brianosman@google.com>
All our other color types put the low bits as the first component in the
name. For 565 B is the low bits and R is the high bits so the name is being
updated to reflect that.
Bug: skia:9170
Change-Id: I67be32440d6c6fa8a345532fe144720d23cf340a
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/221337
Commit-Queue: Greg Daniel <egdaniel@google.com>
Reviewed-by: Robert Phillips <robertphillips@google.com>
This reverts commit edc42b9971.
Reason for revert: Sigh
Original change's description:
> Revert "Revert "Interpreter: Support striped inputs for less overhead""
>
> This reverts commit 645fe10313.
>
> Change-Id: If74a15479f89f49ac33c0b6241bb0db92bc11083
> Reviewed-on: https://skia-review.googlesource.com/c/skia/+/221721
> Reviewed-by: Mike Klein <mtklein@google.com>
> Commit-Queue: Brian Osman <brianosman@google.com>
TBR=mtklein@google.com,brianosman@google.com,reed@google.com
Change-Id: I651d56821bbde8b91887aa885bcf2cb202707388
No-Presubmit: true
No-Tree-Checks: true
No-Try: true
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/221897
Reviewed-by: Brian Osman <brianosman@google.com>
Commit-Queue: Brian Osman <brianosman@google.com>
This is setting up for GrIORefProxy to just become SkRefCnt and GrProxyRef to just become sk_sp.
Change-Id: Ica66565a353de980a7070e0788f1f2b17565baee
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/220297
Commit-Queue: Robert Phillips <robertphillips@google.com>
Reviewed-by: Brian Salomon <bsalomon@google.com>
Change-Id: I6f9ee51f7c063ca03bf48fccd413dae244edd191
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/221778
Reviewed-by: Brian Salomon <bsalomon@google.com>
Commit-Queue: Brian Osman <brianosman@google.com>
This does not need to repeat the code of the version that returns a Node*.
Change-Id: I3c67e3848ba4772cc63b71b78ad29f93fd479956
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/221722
Commit-Queue: Herb Derby <herb@google.com>
Commit-Queue: Mike Klein <mtklein@google.com>
Auto-Submit: Herb Derby <herb@google.com>
Reviewed-by: Mike Klein <mtklein@google.com>
Remove the calling of a deprecated HarfBuzz function. The current use of
HarfBuzz's subsetter calls a deprecated symbol:
hb_subset_input_set_drop_layout(). The change with the deprecation can
be found at [1]. Instead, HarfBuzz disables layout by default, as
indicated by the comment found at [2], rendering the calling of the
function unnecessary in Skia.
[1] 0ff3618c2d (diff-e34a949604c500d3a2aeaafa06cf7d96L142)
[2] https://github.com/harfbuzz/harfbuzz/blob/master/src/hb-subset-input.cc#L55
Bug: chromium:931719
Change-Id: I530e2257d49a6aeb2164bdc2a0358a998d62591f
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/221411
Commit-Queue: Ben Wagner <bungeman@google.com>
Reviewed-by: Hal Canary <halcanary@google.com>
This reverts commit a224fc1105.
Reason for revert: breaking x86-64 bots without AVX2, e.g. Test-Debian9-Clang-GCE-CPU-AVX2-x86_64-Release-All-SK_CPU_LIMIT_SSE2
Original change's description:
> extract Assembler so it can be tested
>
> And start documenting some structs we'll need
> to replace xbyak.
>
> Change-Id: I21c91642799a54e10af85afc8edbe12a9b4aa062
> Reviewed-on: https://skia-review.googlesource.com/c/skia/+/221644
> Reviewed-by: Herb Derby <herb@google.com>
> Commit-Queue: Mike Klein <mtklein@google.com>
TBR=mtklein@google.com,herb@google.com
Change-Id: Ie90d57f66e4d45f94db4ab4f485155533faddae1
No-Presubmit: true
No-Tree-Checks: true
No-Try: true
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/221655
Reviewed-by: Mike Klein <mtklein@google.com>
Commit-Queue: Mike Klein <mtklein@google.com>
This reverts commit 646616a78f.
Reason for revert: Suspected as cause of layout test changes.
Original change's description:
> Consolidate quad optimizations into single internal function.
>
> Routes all non-textured quad draws through single internal function
>
> Change-Id: Ief66864a0ad2d598982c5bf500c8a84ecbf84387
> Reviewed-on: https://skia-review.googlesource.com/c/skia/+/215455
> Commit-Queue: Michael Ludwig <michaelludwig@google.com>
> Reviewed-by: Robert Phillips <robertphillips@google.com>
TBR=robertphillips@google.com,michaelludwig@google.com
Change-Id: I0dc6a0d948c0f5e9221ff6c9fbbbbbb9bc3d9bc0
No-Presubmit: true
No-Tree-Checks: true
No-Try: true
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/221737
Reviewed-by: Brian Salomon <bsalomon@google.com>
Commit-Queue: Brian Salomon <bsalomon@google.com>
And start documenting some structs we'll need
to replace xbyak.
Change-Id: I21c91642799a54e10af85afc8edbe12a9b4aa062
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/221644
Reviewed-by: Herb Derby <herb@google.com>
Commit-Queue: Mike Klein <mtklein@google.com>
This reverts commit 5d89b66ff6.
Reason for revert: Chrome roll
Original change's description:
> Interpreter: Support striped inputs for less overhead
>
> Change-Id: I8c7bd5ed3fb6aebbfb1c5c224acfd73862252621
> Reviewed-on: https://skia-review.googlesource.com/c/skia/+/220778
> Commit-Queue: Brian Osman <brianosman@google.com>
> Reviewed-by: Mike Klein <mtklein@google.com>
> Reviewed-by: Mike Reed <reed@google.com>
TBR=mtklein@google.com,brianosman@google.com,ethannicholas@google.com,reed@google.com
Change-Id: I2275e7ed12031e58e59b6618d0ea59bec2ee2815
No-Presubmit: true
No-Tree-Checks: true
No-Try: true
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/221716
Reviewed-by: Brian Osman <brianosman@google.com>
Commit-Queue: Brian Osman <brianosman@google.com>
Routes all non-textured quad draws through single internal function
Change-Id: Ief66864a0ad2d598982c5bf500c8a84ecbf84387
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/215455
Commit-Queue: Michael Ludwig <michaelludwig@google.com>
Reviewed-by: Robert Phillips <robertphillips@google.com>
We were resetting the vertex buffer state after every draw, even if
the vertex buffer hasn't changed. This addresses that. It also sets
certain renderstate at renderencoder creation -- again, rather than
at every draw.
Bug: skia:8243
Change-Id: I053e150c289c14b60385f555a8b4b771a131ea06
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/221336
Reviewed-by: Greg Daniel <egdaniel@google.com>
Commit-Queue: Jim Van Verth <jvanverth@google.com>
I'm hoping to remove xbyak bit by bit, like the ship of Theseus
or some petrified wood, leaving a functioning homegrown JIT.
This starts with some of the easier instructions, still using
xbyak to accumulate the code buffer, just with db() directly.
- ret
- vzeroupper
- align with noops
Change-Id: I9443bc5191cb1d75b8871d174ccf15ac7e96e9a5
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/221613
Reviewed-by: Herb Derby <herb@google.com>
Commit-Queue: Mike Klein <mtklein@google.com>
This flag is no longer set by any users, so remove it now that baselines
have been updated.
Change-Id: Ib952a93c16a6ca65d89a2009e9732c6339d83f9f
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/221542
Reviewed-by: Herb Derby <herb@google.com>
Commit-Queue: Ben Wagner <bungeman@google.com>
Change-Id: I8c7bd5ed3fb6aebbfb1c5c224acfd73862252621
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/220778
Commit-Queue: Brian Osman <brianosman@google.com>
Reviewed-by: Mike Klein <mtklein@google.com>
Reviewed-by: Mike Reed <reed@google.com>
In GrRecordingContext I moved the auditTrail onto the heap and only there
when compiling for tests. This allowed us to move a lot of files out of
include private.
Change-Id: Ib76ac211c0c6fd10bacaccf0c5f93f21a59f35d5
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/221344
Commit-Queue: Greg Daniel <egdaniel@google.com>
Reviewed-by: Brian Salomon <bsalomon@google.com>
Auto-Submit: Greg Daniel <egdaniel@google.com>
This reverts commit 5509102043.
Reason for revert: valgrind issues
Original change's description:
> Add function to GrDataUtils to handle color conversions.
>
> Like SkConvertPixels but knows about all GrColorTypes, origin, and can
> apply an arbitrary GrSwizzle.
>
> Use in GrSurfaceContext read/write pixels methods.
>
> Add support for '0' to GrSwizzle.
>
>
> Change-Id: Ib9dd215fcb0ee8b33c4020893c22b4ab7ce1f40b
> Reviewed-on: https://skia-review.googlesource.com/c/skia/+/220761
> Commit-Queue: Brian Salomon <bsalomon@google.com>
> Reviewed-by: Greg Daniel <egdaniel@google.com>
TBR=egdaniel@google.com,bsalomon@google.com,brianosman@google.com
Change-Id: If50f3e26875787d9309009e9c701774fbad0afda
No-Presubmit: true
No-Tree-Checks: true
No-Try: true
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/221538
Reviewed-by: Robert Phillips <robertphillips@google.com>
Commit-Queue: Robert Phillips <robertphillips@google.com>
This reverts commit 2935638a1b.
Reason for revert: valgrind errors
Original change's description:
> Make SkASSERT not break build in constexpr GrSwizzle method on GCC
>
> Change-Id: I042b90e7c405505447662e6d187ca1519efd4743
> Reviewed-on: https://skia-review.googlesource.com/c/skia/+/221342
> Reviewed-by: Brian Osman <brianosman@google.com>
> Commit-Queue: Brian Salomon <bsalomon@google.com>
TBR=egdaniel@google.com,bsalomon@google.com,brianosman@google.com
Change-Id: I6bc0f8f84212e3ce53f4d2198c4369db7f08ea3f
No-Presubmit: true
No-Tree-Checks: true
No-Try: true
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/221537
Reviewed-by: Robert Phillips <robertphillips@google.com>
Commit-Queue: Robert Phillips <robertphillips@google.com>
This reverts commit 90507286cc.
Reason for revert: Seems to be breaking some builds
Original change's description:
> Shuffle SkSL sources around so compiler and bytecode can be used w/o GPU
>
> Change-Id: I7236a30040ab532086e68d6e9de2898dd7acaa32
> Reviewed-on: https://skia-review.googlesource.com/c/skia/+/221098
> Commit-Queue: Brian Osman <brianosman@google.com>
> Reviewed-by: Mike Reed <reed@google.com>
> Reviewed-by: Mike Klein <mtklein@google.com>
TBR=mtklein@google.com,kjlubick@google.com,brianosman@google.com,ethannicholas@google.com,reed@google.com
Change-Id: Ie230315a72ebcfae32bc9ce7bafec1f87106cff2
No-Presubmit: true
No-Tree-Checks: true
No-Try: true
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/221536
Reviewed-by: Robert Phillips <robertphillips@google.com>
Commit-Queue: Robert Phillips <robertphillips@google.com>
The passed localeName and defaultFamilyName and associated lengths are
expected to include the terminating NULL (mostly due to the way LOGFONT
and GetUserDefaultLocaleName are defined). This implies that in the case
where these values cannot be retrieved the empty string should be
passed.
Bug: chromium:973361
Change-Id: I0ac98f5250e1cb2d7fa426b255c5e2c2619dc51c
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/221417
Commit-Queue: Ben Wagner <bungeman@google.com>
Reviewed-by: Mike Reed <reed@google.com>
Reviewed-by: Herb Derby <herb@google.com>
Remove the SkBaseMutex (and SkBaseSemaphore). This allows all the thread
annotation machinery to work.
Change-Id: I2da420ec3165ccbcd90c474c0b62bfef42df2a53
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/221340
Commit-Queue: Herb Derby <herb@google.com>
Reviewed-by: Mike Klein <mtklein@google.com>
We've got slightly different implementations of unaligned_load(), and
that's causing a x86 Chromium Debug builds to go a bit haywire. A
single best implementation with an SK_ALWAYS_INLINE and explanation
should help things.
Might as well move its companion unaligned_store() too.
Bug: chromium:974542
Change-Id: If9f3eb4d33bfb8390c661f9e196122da0b9b84ca
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/221409
Reviewed-by: Nico Weber <thakis@chromium.org>
Commit-Queue: Mike Klein <mtklein@google.com>
This new cap tells Ganesh how many samples to use when performing
internal draws with MSAA or mixed samples. The default is always 4x,
but the client can change that with
GrContextOptions::fPreferredInternalMSAASampleCount.
Also adds a command line flag to viewer to control
fPreferredInternalMSAASampleCount.
Bug: skia:
Change-Id: Iba369273e802aa1bee796b576b3c18af347b0494
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/221156
Commit-Queue: Chris Dalton <csmartdalton@google.com>
Reviewed-by: Brian Salomon <bsalomon@google.com>
--motion_angle ... [default is 180]
--motion_samples ... [default is 1, for no motion blur]
Change-Id: Iec0f31655b3369f51e0b398efb2d5b156dcbaf2e
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/221416
Reviewed-by: Mike Reed <reed@google.com>
Commit-Queue: Mike Reed <reed@google.com>
Auto-Submit: Mike Reed <reed@google.com>
Change-Id: I7236a30040ab532086e68d6e9de2898dd7acaa32
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/221098
Commit-Queue: Brian Osman <brianosman@google.com>
Reviewed-by: Mike Reed <reed@google.com>
Reviewed-by: Mike Klein <mtklein@google.com>
Change-Id: I042b90e7c405505447662e6d187ca1519efd4743
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/221342
Reviewed-by: Brian Osman <brianosman@google.com>
Commit-Queue: Brian Salomon <bsalomon@google.com>
Like SkConvertPixels but knows about all GrColorTypes, origin, and can
apply an arbitrary GrSwizzle.
Use in GrSurfaceContext read/write pixels methods.
Add support for '0' to GrSwizzle.
Change-Id: Ib9dd215fcb0ee8b33c4020893c22b4ab7ce1f40b
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/220761
Commit-Queue: Brian Salomon <bsalomon@google.com>
Reviewed-by: Greg Daniel <egdaniel@google.com>
Make sure we're using five or fewer arguments.
Today all programs use one or two arguments, so this doesn't
really have any immediate effect, but it should be there.
Change-Id: Ia85e56ef63ceb442702546c402cd11a13daa2c25
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/221270
Reviewed-by: Mike Klein <mtklein@google.com>
Commit-Queue: Mike Klein <mtklein@google.com>
I'm staring at this assembly,
vmovups (%rsi), %ymm3
vpsrld $24, %ymm3, %ymm4
vpslld $16, %ymm4, %ymm15
vorps %ymm4, %ymm15, %ymm4
vpsubw %ymm4, %ymm0, %ymm4
Just knowing that could be
vmovups (%rsi), %ymm3
vpshufb 0x??(%rip), %ymm3, %ymm4
vpsubw %ymm4, %ymm0, %ymm4
That is, instead of shifting, shifting, and bit-oring
to create the 0a0a scale factor from ymm3, we could just
byte shuffle directly using some pre-baked control pattern
(stored at the end of the program like other constants)
pshufb lets you arbitrarily remix bytes from its argument and
zero bytes, and NEON has a similar family of vtbl instructions,
even including that same feature of injecting zeroes.
I think I've got this working, and the speedup is great,
from 0.19 to 0.16 ns/px for I32_SWAR, and
from 0.43 to 0.38 ns/px for I32.
Change-Id: Iab850275e826b4187f0efc9495a4b9eab4402c38
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/220871
Commit-Queue: Mike Klein <mtklein@google.com>
Reviewed-by: Herb Derby <herb@google.com>
With the modified gm, the time (on imac pro) goes from 2.4 to 1.6
Change-Id: I9f940220c129f74771f3b17126657bcf3739044f
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/221176
Commit-Queue: Mike Reed <reed@google.com>
Reviewed-by: Mike Klein <mtklein@google.com>
Reviewed-by: Ethan Nicholas <ethannicholas@google.com>
Change-Id: I20e652f2b6f9bf606b03c6dd4e346c3439ea8a0b
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/220876
Reviewed-by: Brian Osman <brianosman@google.com>
Commit-Queue: Herb Derby <herb@google.com>
Change-Id: I66713976f08b1dbf0966d9a901f666b9f834b659
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/221096
Reviewed-by: Brian Osman <brianosman@google.com>
Commit-Queue: Brian Osman <brianosman@google.com>
Slightly sharper, but far easier to hold:
- Remove Value union from interface, everything is a 32-bit
value type, or a collection thereof.
- Collapse to one version of Run (that takes count), and make
it a member on ByteCode.
- Similarly, move disassemble to ByteCodeFunction.
Change-Id: I07c85e65991178b3f52e20e815c25f36bc9c4257
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/220753
Commit-Queue: Brian Osman <brianosman@google.com>
Reviewed-by: Mike Klein <mtklein@google.com>
