SkRRectPriv::IsSimpleCircular does a tolerance-based check for the
rrect radii but not for determining if the rrect is simple. This adds
IsNearlySimpleCircular as a looser check, since we don't care in this
case that the rrect is perfectly simple.
Bug: skia:11717
Change-Id: Ibbd53ff8c120287339fa29f87f064e6fb633ee7c
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/382703
Reviewed-by: Leon Scroggins <scroggo@google.com>
Commit-Queue: Jim Van Verth <jvanverth@google.com>
This crossed in the mail with http://review.skia.org/382476
Change-Id: I09ca2ff20bb2dc0846b591893d06823577faea29
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/382758
Commit-Queue: John Stiles <johnstiles@google.com>
Commit-Queue: Chris Dalton <csmartdalton@google.com>
Auto-Submit: John Stiles <johnstiles@google.com>
Reviewed-by: Chris Dalton <csmartdalton@google.com>
Change-Id: I3edc1812f74319abeb13d94cf7e2426bd68b57a3
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/382282
Commit-Queue: Jim Van Verth <jvanverth@google.com>
Reviewed-by: Robert Phillips <robertphillips@google.com>
For things like path rendering we can (and should) do better than an
atlas. But for use cases like "many small paths clipped by many small
paths," an atlas is probably still the best solution we have.
Change-Id: Iff5d250d0d6ccba9235f3fe652891554e3fdfc24
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/380856
Reviewed-by: Michael Ludwig <michaelludwig@google.com>
Commit-Queue: Chris Dalton <csmartdalton@google.com>
This is useful because it allows the variables to be declared as `const`
when they are trivial values like `half4(1)`. This enables the constant
folder to simplify or eliminate them. In most cases, this is only a
small benefit, as you'd expect a competent GPU driver to do the same.
However, Mali-400 can benefit significantly from optimizing away the
multiplication against a constant half4(1) coverage in Porter-Duff.
Mali-400 performance is back to normal: http://screen/3cDxdaGkYE8oBcS
Change-Id: I21fd23f91f747079cd05b082f7b3444aeabafb93
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/382476
Auto-Submit: John Stiles <johnstiles@google.com>
Commit-Queue: Ethan Nicholas <ethannicholas@google.com>
Reviewed-by: Ethan Nicholas <ethannicholas@google.com>
This block no longer appears to provide any benefit; we mangle variable
names to prevent name collisions already.
Change-Id: I970a9d55ed5a77b93ff611e6a58e9424db78f6db
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/382396
Auto-Submit: John Stiles <johnstiles@google.com>
Reviewed-by: Ethan Nicholas <ethannicholas@google.com>
Commit-Queue: John Stiles <johnstiles@google.com>
When optimization is disabled, I did not expect to see arithmetic like
`color * ONE` disappear (even though it's just a constant fold):
const vec4 ONE = half4(1);
vec4 x = color * ONE;
Change-Id: Id24a632712591c73a904b90e5efdba092d846c0e
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/382477
Commit-Queue: Ethan Nicholas <ethannicholas@google.com>
Auto-Submit: John Stiles <johnstiles@google.com>
Reviewed-by: Ethan Nicholas <ethannicholas@google.com>
As you might expect, a function tagged with `noinline` will never be
considered as a candidate for inlining.
Change-Id: Ia098f8974e6de251d78bb2a76cd71db8a86bc19c
Bug: skia:11362
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/382337
Reviewed-by: Ethan Nicholas <ethannicholas@google.com>
Commit-Queue: Ethan Nicholas <ethannicholas@google.com>
Auto-Submit: John Stiles <johnstiles@google.com>
This reverts commit dc3d678712.
Reason for revert: Crashing on exit thread handler
Original change's description:
> Reland "cache the creation of one GrAtlasTextOp"
>
> This is a reland of 4b1fb7ca90
>
> Original change's description:
> > cache the creation of one GrAtlasTextOp
> >
> > GrAtlasTextOp has a high probability of being merged with the
> > previous op. This cache keeps using the same op to merge with
> > keeping memory warm.
> >
> > This show about 5.75% improvement in skpbench on desk_nytimes.
> >
> > When compiling for ios 9 or earlier, this optimization is
> > disabled.
> >
> > Change-Id: I13ccbef6dcd4b9d82103bf20bba7d94f3e4fb6f4
> > Reviewed-on: https://skia-review.googlesource.com/c/skia/+/376718
> > Reviewed-by: Michael Ludwig <michaelludwig@google.com>
> > Reviewed-by: Brian Salomon <bsalomon@google.com>
> > Commit-Queue: Herb Derby <herb@google.com>
>
> Change-Id: I935a2965062b1fddb28806e85eb0fe055ba46ec2
> Reviewed-on: https://skia-review.googlesource.com/c/skia/+/380320
> Commit-Queue: Herb Derby <herb@google.com>
> Reviewed-by: Michael Ludwig <michaelludwig@google.com>
TBR=bsalomon@google.com,herb@google.com,michaelludwig@google.com
Change-Id: I3bc3329580460fcf8c0b49f655a88cb902da3d58
No-Presubmit: true
No-Tree-Checks: true
No-Try: true
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/382556
Reviewed-by: Herb Derby <herb@google.com>
Commit-Queue: Herb Derby <herb@google.com>
This is a reland of 4b1fb7ca90
Original change's description:
> cache the creation of one GrAtlasTextOp
>
> GrAtlasTextOp has a high probability of being merged with the
> previous op. This cache keeps using the same op to merge with
> keeping memory warm.
>
> This show about 5.75% improvement in skpbench on desk_nytimes.
>
> When compiling for ios 9 or earlier, this optimization is
> disabled.
>
> Change-Id: I13ccbef6dcd4b9d82103bf20bba7d94f3e4fb6f4
> Reviewed-on: https://skia-review.googlesource.com/c/skia/+/376718
> Reviewed-by: Michael Ludwig <michaelludwig@google.com>
> Reviewed-by: Brian Salomon <bsalomon@google.com>
> Commit-Queue: Herb Derby <herb@google.com>
Change-Id: I935a2965062b1fddb28806e85eb0fe055ba46ec2
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/380320
Commit-Queue: Herb Derby <herb@google.com>
Reviewed-by: Michael Ludwig <michaelludwig@google.com>
With skia-review.googlesource.com/c/skia/+/363782, both cpu and gpu
backends would gracefully fail readPixel requests where dst row bytes
wasn't a multiple of dst bpp. It also updated the cpu backend's
writePixels behavior to gracefully reject writePixels requests where
the src row bytes wasn't a multiple of src bpp.
GPU writePixels would not detect this and later trigger an assert
in debug builds in GrConvertPixels (caught by the linked fuzzer bug).
This adds tests to mirror the read pixels bad-row-bytes tests and
updates GrSurfaceContext::writePixels to check src row bytes vs. bpp.
I confirmed it fixes the fuzzer crash.
Bug: chromium:1185266
Change-Id: I7cd8406c65a9ba35a55d695b2f65410a1edd2a19
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/382276
Auto-Submit: Michael Ludwig <michaelludwig@google.com>
Commit-Queue: Brian Salomon <bsalomon@google.com>
Reviewed-by: Brian Salomon <bsalomon@google.com>
Acts similarly to the GL precompiled shaders, in that it will cache
preliminary data (in this case, the MTLLibraries) in the PipelineState
cache, and then when the pipeline is built will extract that. Once
the pipeline is built the MTLLibraries are discarded.
Bug: skia:11392
Change-Id: I701e2e857d105c9149a2f69fe7b60bf4599195bc
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/380457
Reviewed-by: Brian Osman <brianosman@google.com>
Commit-Queue: Jim Van Verth <jvanverth@google.com>
Change-Id: Ic2d1240ab785101365b0fd934562505fb5a3e599
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/381816
Reviewed-by: Brian Osman <brianosman@google.com>
Commit-Queue: Brian Osman <brianosman@google.com>
Auto-Submit: John Stiles <johnstiles@google.com>
It turned out that everywhere we were using or testing DSL code either
directly or indirectly imported big chunks of the SkSL library. These
imports turned out to be necessary; code written using just DSL.h would
fail with various template instantiation errors.
Change-Id: Iae72d15b0d6ef14614ac1a4ff08c36bc1876cd4d
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/381638
Commit-Queue: Ethan Nicholas <ethannicholas@google.com>
Reviewed-by: John Stiles <johnstiles@google.com>
We should avoid convenience factory methods when possible, to
keep things out of the autorelease pool.
Change-Id: I0a042b59619aff7a91b889add4ad341ab892809a
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/381860
Auto-Submit: Adlai Holler <adlai@google.com>
Commit-Queue: Jim Van Verth <jvanverth@google.com>
Reviewed-by: Jim Van Verth <jvanverth@google.com>
The crash is because method GrGpuBuffer::onRelease() is called on wrong
thread. SkMessageBus::Post(const Message& m) takes a const ref of the
message, so there is a little possibility the GPU thread received and
handled the message before SkMessageBus::Post() is returned. In that
case, the caller of SkMessageBus::Post() (AsyncReadResult) still holds
the last ref of the GrGpuBuffer, and then GrGpuBuffer() will just be
released on the wrong thread.
Bug: chromium:1185489
Change-Id: I28665dbb1db7925d59ec574e9e26385e845ff4df
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/380696
Commit-Queue: Peng Huang <penghuang@chromium.org>
Commit-Queue: Brian Salomon <bsalomon@google.com>
Reviewed-by: Brian Salomon <bsalomon@google.com>
Auto-Submit: Peng Huang <penghuang@chromium.org>
This is sort of a nuclear option for matrices that are archived by e.g.
SkPictures, SkShaders, and SkEffects. But in all these cases, the
unarchived objects are const and ought to be thread-safe.
Bug: skia:11715
Change-Id: I0215f0fc988e9ce01efd4e200c9951c3b5be7da6
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/381519
Commit-Queue: Adlai Holler <adlai@google.com>
Reviewed-by: Robert Phillips <robertphillips@google.com>
Reviewed-by: Mike Reed <reed@google.com>
We can have a stencil attachment on the RenderTarget with stencil disabled.
Bug: skia:11392
Change-Id: Iefc661e9206eb859b02e91857f3682a525493df6
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/380716
Reviewed-by: Brian Osman <brianosman@google.com>
Commit-Queue: Jim Van Verth <jvanverth@google.com>
These were accidentally omitted from the supported operator list.
Change-Id: Idecd17adb8b3f5043e36328c65ca12be33e990f4
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/381637
Auto-Submit: Ethan Nicholas <ethannicholas@google.com>
Reviewed-by: John Stiles <johnstiles@google.com>
Failed instantiations are ultra-rare, so instead of trying to return
early in that case, go ahead and finish iterating through the list
so that it's cleared out and the cleanup-assertions in the
destructor pass.
Bug: skia:10877
Change-Id: I1452e4521fb35a374a046e15df1fe33630f72df9
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/381756
Auto-Submit: Adlai Holler <adlai@google.com>
Reviewed-by: Chris Dalton <csmartdalton@google.com>
Commit-Queue: Adlai Holler <adlai@google.com>
Expressions like `value == value` or `color.a != color.a` can be
replaced by `true` or `false` on sight. The GLSL spec makes it clear
that checking for NaN is optional:
4.7.1 Range and Precision
"... NaNs are not required to be generated. Support for signaling NaNs
is not required and exceptions are never raised. Operations and built-in
functions that operate on a NaN are not required to return a NaN as the
result."
Change-Id: I2ad9f2dc505b638ea2904bef41b7a79a2b329551
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/381262
Commit-Queue: John Stiles <johnstiles@google.com>
Auto-Submit: John Stiles <johnstiles@google.com>
Reviewed-by: Brian Osman <brianosman@google.com>
This CL will be used to test for potential performance regressions (or
improvements) that we might cause by disabling this optimization pass.
It will be reverted in ~1 day.
Change-Id: I26b7687c341eb6d81231406381c39869cfccf6d6
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/381259
Auto-Submit: John Stiles <johnstiles@google.com>
Commit-Queue: John Stiles <johnstiles@google.com>
Reviewed-by: Ethan Nicholas <ethannicholas@google.com>
Reviewed-by: Brian Osman <brianosman@google.com>
Change-Id: I20e4f2168713cbd54c0428c60e38a736cf735df3
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/381476
Reviewed-by: Brian Osman <brianosman@google.com>
Commit-Queue: Herb Derby <herb@google.com>
This can cause unreachable-code warnings when control-flow analysis is
enabled, and CFA has a matching optimization pass, so we only do this
when CFA is turned off.
Change-Id: I6c90bbcc0a9de2478a22dd9054567dca1b13b84a
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/381258
Commit-Queue: John Stiles <johnstiles@google.com>
Commit-Queue: Ethan Nicholas <ethannicholas@google.com>
Auto-Submit: John Stiles <johnstiles@google.com>
Reviewed-by: Ethan Nicholas <ethannicholas@google.com>
Reviewed-by: Brian Osman <brianosman@google.com>
There's really only one failure at this point, and it's a failed
instantiation. The original intention of the allocation failure
system was to only drop the ops that referred to the uninstantiated
proxies, but somewhere along the way that was lost and
we started dropping arbitrarily large chunks of ops. Lets just
bail and not crash instead.
Bug: skia:10877
Change-Id: I675358e8a1fbd2d75ea29b72ccfc50c7df90343e
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/371337
Commit-Queue: Adlai Holler <adlai@google.com>
Reviewed-by: Robert Phillips <robertphillips@google.com>
Use placement-new to copy AST data into fBytes; this gives us a "real
object" at that memory location which we can later access via
reinterpret_cast. In practical terms, this lets us return a reference
that points inside of fBytes without causing UB. This saves us from
needing to copy the data back out every time it is accessed.
For large objects, this can be a substantial savings. For sufficiently
small objects (<= 8 bytes) the data is copied back out regardless, since
a single load is more than fast enough.
Change-Id: I3ad246acda4feb7e7aa1987d966abd607983e205
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/380996
Commit-Queue: John Stiles <johnstiles@google.com>
Auto-Submit: John Stiles <johnstiles@google.com>
Reviewed-by: Brian Osman <brianosman@google.com>
This lets us see at a glance what each optimization pass contributes to
our final compiled SkSL output.
Change-Id: I52c56c92c408eee34045c5e6f60298cf9548ff5d
Bug: skia:11319
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/381257
Commit-Queue: John Stiles <johnstiles@google.com>
Commit-Queue: Ethan Nicholas <ethannicholas@google.com>
Auto-Submit: John Stiles <johnstiles@google.com>
Reviewed-by: Ethan Nicholas <ethannicholas@google.com>
This makes the filter thread-safe.
Bug: skia:10286
Change-Id: I0d8796655bd51f21fe7cb30f02340b87cb13c665
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/381299
Commit-Queue: Adlai Holler <adlai@google.com>
Commit-Queue: Brian Osman <brianosman@google.com>
Auto-Submit: Adlai Holler <adlai@google.com>
Reviewed-by: Brian Osman <brianosman@google.com>
Change-Id: I7a7874e58bf53978afce8a41b26092406b6490ed
Bug: skia:11342
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/380360
Reviewed-by: Brian Osman <brianosman@google.com>
Commit-Queue: John Stiles <johnstiles@google.com>
findProxyByUniqueKey merely checks if a proxy exists for a key while
findOrCreateProxyByUniqueKey checks if there is an existing texture
whose proxy was deleted and recreates a new proxy for it.
Change-Id: I352639d03769dbb9659819d52bc6927347b5a452
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/381222
Auto-Submit: Brian Salomon <bsalomon@google.com>
Commit-Queue: Mike Reed <reed@google.com>
Reviewed-by: Mike Reed <reed@google.com>
This reverts commit 9ef3f2e3da.
Reason for revert: relanding with fix
Original change's description:
> Revert "Have GrVkRenderTarget only use GrVkAttachments and not derive from GrVkImage."
>
> This reverts commit 3dc6c190da.
>
> Reason for revert: hitting assert about RT having input attachment on mali bots
>
> Original change's description:
> > Have GrVkRenderTarget only use GrVkAttachments and not derive from GrVkImage.
> >
> > This change moves the color and resolve attachments used in a
> > GrVkRenderTarget to be a GrVkAttachment. These along with the msaa
> > attachment now mean that GrVkRenderTarget no longer needs to derive from
> > a GrVkImage.
> >
> > There are a couple ugly things in this CL since GrVkTexture still is a
> > GrVkImage since we can't share attachments between GrVkRT and GrVkTex.
> > But when that gets updated in the follow on CL things will look much nicer.
> >
> > Bug: skia:10727
> > Change-Id: I2f12674d7517c6d6dea389e2d1fb7296028bcc85
> > Reviewed-on: https://skia-review.googlesource.com/c/skia/+/379576
> > Reviewed-by: Brian Salomon <bsalomon@google.com>
> > Commit-Queue: Greg Daniel <egdaniel@google.com>
>
> TBR=egdaniel@google.com,jvanverth@google.com,bsalomon@google.com
>
> Change-Id: Ic46f3947ed9f7b2ca26e8418d643e7f89b6108d2
> No-Presubmit: true
> No-Tree-Checks: true
> No-Try: true
> Bug: skia:10727
> Reviewed-on: https://skia-review.googlesource.com/c/skia/+/380459
> Reviewed-by: Greg Daniel <egdaniel@google.com>
> Commit-Queue: Greg Daniel <egdaniel@google.com>
# Not skipping CQ checks because this is a reland.
Bug: skia:10727
Change-Id: I7a995ee9ad35bdac34cfcfd6b0d963c3e0bb90b8
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/380460
Reviewed-by: Greg Daniel <egdaniel@google.com>
Commit-Queue: Greg Daniel <egdaniel@google.com>
Expressions like `x * 1`, `x *= 1`, `x + 0`, `x * 0`, or `0 / x` don't
actually do anything, and can be simplified to just `x` or `0`. (The
zero case must also check that `x` doesn't have side effects, because
`0 * myFunction()` still needs to call `myFunction`.)
`0 - x` is also detected and rewritten as `-x`.
`0 / 0` is left as-is.
This logic works for scalars and vectors; matrices are left as-is.
A similar optimization also occurs in the constant-propagation pass, so
we see almost no diffs in the tests. If control-flow analysis is turned
off, we do see some improvements. (I didn't reuse the existing code at
all, since it was designed around rewriting the CFG tree, but the
concept was identical.)
Change-Id: Ia99cd81f1d4cd3dafaa43ccac6a2261e3257a185
Bug: skia:11343
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/380356
Reviewed-by: Brian Osman <brianosman@google.com>
Commit-Queue: John Stiles <johnstiles@google.com>
Auto-Submit: John Stiles <johnstiles@google.com>
This is a reland of cad44bcdd2
Original change's description:
> Use analytic AA in FillRRectOp even if the target has MSAA
>
> Bug: skia:11396
> Change-Id: Ia29fd5c9f00006274bb2aa1db96fbad57d858786
> Reviewed-on: https://skia-review.googlesource.com/c/skia/+/380016
> Commit-Queue: Chris Dalton <csmartdalton@google.com>
> Reviewed-by: Robert Phillips <robertphillips@google.com>
Bug: skia:11396
Change-Id: I6382085ccd037b534a1ed1355c1464aaef799a44
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/380477
Reviewed-by: Robert Phillips <robertphillips@google.com>
Commit-Queue: Chris Dalton <csmartdalton@google.com>
This reverts commit 3dc6c190da.
Reason for revert: hitting assert about RT having input attachment on mali bots
Original change's description:
> Have GrVkRenderTarget only use GrVkAttachments and not derive from GrVkImage.
>
> This change moves the color and resolve attachments used in a
> GrVkRenderTarget to be a GrVkAttachment. These along with the msaa
> attachment now mean that GrVkRenderTarget no longer needs to derive from
> a GrVkImage.
>
> There are a couple ugly things in this CL since GrVkTexture still is a
> GrVkImage since we can't share attachments between GrVkRT and GrVkTex.
> But when that gets updated in the follow on CL things will look much nicer.
>
> Bug: skia:10727
> Change-Id: I2f12674d7517c6d6dea389e2d1fb7296028bcc85
> Reviewed-on: https://skia-review.googlesource.com/c/skia/+/379576
> Reviewed-by: Brian Salomon <bsalomon@google.com>
> Commit-Queue: Greg Daniel <egdaniel@google.com>
TBR=egdaniel@google.com,jvanverth@google.com,bsalomon@google.com
Change-Id: Ic46f3947ed9f7b2ca26e8418d643e7f89b6108d2
No-Presubmit: true
No-Tree-Checks: true
No-Try: true
Bug: skia:10727
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/380459
Reviewed-by: Greg Daniel <egdaniel@google.com>
Commit-Queue: Greg Daniel <egdaniel@google.com>
This gives Viewer a way to dynamically toggle CFA off and on, so we can
immediately see the impact on shader optimization.
Change-Id: I912df70de126a74a1f0d11c511b2acc64f1e9f28
Bug: skia:11319
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/380456
Commit-Queue: John Stiles <johnstiles@google.com>
Auto-Submit: John Stiles <johnstiles@google.com>
Reviewed-by: Ethan Nicholas <ethannicholas@google.com>
Declare
This solves several issues caused by the lack of ordering guarantees in
C++; it was possible for the SkSL backend to look for the value of a
variable before its Declare() call gets processed. Moving the initial
value out of Declare should fix this whole class of problems.
Change-Id: I428fe230f1c312a0128c1f00c2a36cb95f4590a6
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/380358
Reviewed-by: John Stiles <johnstiles@google.com>
Reviewed-by: Brian Osman <brianosman@google.com>
This reverts commit cad44bcdd2.
Reason for revert: Triggered an Adredo330 bug
Original change's description:
> Use analytic AA in FillRRectOp even if the target has MSAA
>
> Bug: skia:11396
> Change-Id: Ia29fd5c9f00006274bb2aa1db96fbad57d858786
> Reviewed-on: https://skia-review.googlesource.com/c/skia/+/380016
> Commit-Queue: Chris Dalton <csmartdalton@google.com>
> Reviewed-by: Robert Phillips <robertphillips@google.com>
TBR=robertphillips@google.com,csmartdalton@google.com
Change-Id: Ic51a279f9fefe03b28182bf1222c2e26625f5bf3
No-Presubmit: true
No-Tree-Checks: true
No-Try: true
Bug: skia:11396
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/380476
Reviewed-by: Chris Dalton <csmartdalton@google.com>
Commit-Queue: Chris Dalton <csmartdalton@google.com>
Whenever we need the description (as we're putting something in the
persistent cache), go back and re-compute the key to get the
description. This pattern means that we pay a (very) small overhead for
virtual dispatch in the common case, and only pay for the expensive
string work when we're already doing expensive work (compiling).
Bug: skia:11372
Change-Id: I3d4dc19f2d8883f8117f5f6489fc852cf9503eb3
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/380359
Reviewed-by: Brian Salomon <bsalomon@google.com>
Commit-Queue: Brian Osman <brianosman@google.com>
- Values from constant variables are folded in when it helps, e.g.:
const bool SHINY = true;
const float SHININESS = 2;
if (!SHINY) { // <-- optimizes directly to `false`
param = -SHININESS; // <-- optimizes to `-2`
- Doubled-up logical-not and negation are stripped:
y = -(-x); // <-- optimizes to `y = x;`
b = !!a; // <-- optimizes to `b = a;`
Removal of doubled-up negation and logical-not was actually never
implemented in the constant-propagation phase; I just noticed it while
I was here and thinking about it.
Change-Id: Ie28bb9b5af91376f03d926e26e37f4a131bbf550
Bug: skia:11343
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/379298
Auto-Submit: John Stiles <johnstiles@google.com>
Commit-Queue: John Stiles <johnstiles@google.com>
Reviewed-by: Brian Osman <brianosman@google.com>
This enables the ternary to be optimized away in code like:
const bool SHINY = true;
color = SHINY ? add_shine(x) : x; // to --> `color = add_shine(x);`
Without constant propagation.
Also, I added a unit test for ternary expression simplification; I
wasn't able to find an existing one.
When the optimization flag is disabled, this CL actually removes the
optimization of `true ? x : y` --> `x` entirely; previously, this
substitution would be made regardless of optimization settings.
Change-Id: I93a8b9d4027902d35f8a19cfd6417170b209d056
Bug: skia:11343
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/379297
Reviewed-by: Brian Osman <brianosman@google.com>
Commit-Queue: John Stiles <johnstiles@google.com>
Auto-Submit: John Stiles <johnstiles@google.com>
This change moves the color and resolve attachments used in a
GrVkRenderTarget to be a GrVkAttachment. These along with the msaa
attachment now mean that GrVkRenderTarget no longer needs to derive from
a GrVkImage.
There are a couple ugly things in this CL since GrVkTexture still is a
GrVkImage since we can't share attachments between GrVkRT and GrVkTex.
But when that gets updated in the follow on CL things will look much nicer.
Bug: skia:10727
Change-Id: I2f12674d7517c6d6dea389e2d1fb7296028bcc85
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/379576
Reviewed-by: Brian Salomon <bsalomon@google.com>
Commit-Queue: Greg Daniel <egdaniel@google.com>
Swizzle optimizations now occur at IR generation time. These
optimizations are redundant with the control-flow optimization phase so
they are mostly not visible in our test output, but they do affect DSL
test results. Interestingly, they do improve our test output slightly
as well, for various reasons (e.g. we do not fully optimize lvalues in
the control-flow pass).
Change-Id: I6ebe6d71a5c22d9823b5fa500e43078915cbfb45
Bug: skia:11343
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/372257
Reviewed-by: Brian Osman <brianosman@google.com>
Commit-Queue: Brian Osman <brianosman@google.com>
Auto-Submit: John Stiles <johnstiles@google.com>
Makes SkImage_Gpu backed by two proxies, an original and a copy. The
image uses the original until a new render task is bound to it at which
point further uses of the image will use the copy. If the image is ever
used off a GrDirectContext we fall over to the copy. If the copy is
never used and never can be used by the next flush then the render
task that populates it is marked "skipped" and we don't perform the
copy.
Bug: skia:11208
Change-Id: Id255f4a733acc608c8a53c1a5633207aeafc404b
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/366282
Commit-Queue: Brian Salomon <bsalomon@google.com>
Reviewed-by: Robert Phillips <robertphillips@google.com>
This is a reland of 40a9061203
Original change's description:
> rename GrSDFTOptions to GrSDFTControl
>
> Change-Id: Ie03fce7a99a9f71b18d54e3cd35e7675fb7f8912
> Reviewed-on: https://skia-review.googlesource.com/c/skia/+/379616
> Commit-Queue: Herb Derby <herb@google.com>
> Reviewed-by: Robert Phillips <robertphillips@google.com>
Change-Id: Iba0816159a8ef99448a0040c487e56700b96be5d
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/379845
Reviewed-by: Robert Phillips <robertphillips@google.com>
Commit-Queue: Herb Derby <herb@google.com>
