Adds loads/stores (but not dst loads/gathers) for:
*fp16 single and two channel
*fp32 two channel
*normalized unsigned 8 bit two channel
*normalized unsigned 16 bit single and two channel
TODO: 4 channel unsigned normalized 16 bit.
MISSING: fp32 single channel. No matching current or future GrColorType
planned AFAIK.
Intent is to support all (noncompressed) GrColorType load/stores in
order to implement fallbacks for 3D API limitations on texture uploads
and render target readbacks (some of which require swizzling).
Also, can be used to support YUV<->RGB planar
splitters/joiners on CPU.
Bug: skia:8962
Change-Id: I258d5682a1f4025b31639a97b1a1a02077a2453f
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/219999
Reviewed-by: Mike Klein <mtklein@google.com>
Commit-Queue: Brian Salomon <bsalomon@google.com>
This CL leaves the Ganesh world in a very odd place. In particular it still:
has pendingIO refs on GrSurfaces
forwards the proxy refs on to the backing GrSurface
Removing everything at once only ends in a mess thus, this goofball CL.
Change-Id: If112ff311bcef2e8d65a36c3b53b0ded4041c24e
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/210040
Commit-Queue: Robert Phillips <robertphillips@google.com>
Reviewed-by: Brian Salomon <bsalomon@google.com>
Remove the need to include SkTypes.h in SkVx.h,
making SkVx entirely independent of Skia.
As an experiment, switch to checking Clang/GCC-style __SSE__ /
__ARM_NEON defines directly instead of the slightly more abstract
SK_CPU_SSE_LEVEL / SK_ARM_HAS_NEON.
Those SK_ defines only exist to help SSE detection on MSVC, which SkVx
generates serial code for anyway.
If this sticks I may do this same sort of change all through Skia.
Change-Id: I1c51fd6ba1fa48f199ce623824d5ef20ff6be995
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/219541
Reviewed-by: Brian Osman <brianosman@google.com>
Reviewed-by: Michael Ludwig <michaelludwig@google.com>
Commit-Queue: Mike Klein <mtklein@google.com>
This reverts commit 4c6f9b7670.
Reason for revert: Landing with neuxs 7 and androind one fixes
Original change's description:
> Revert "Reland "Remove support for copyAsDraw in gpu copySurface.""
>
> This reverts commit 84ea04949c.
>
> Reason for revert: nexus 7 and android one broken
>
> Original change's description:
> > Reland "Remove support for copyAsDraw in gpu copySurface."
> >
> > This reverts commit c5167c053b.
> >
> > Reason for revert: fixed
> >
> > Original change's description:
> > > Revert "Remove support for copyAsDraw in gpu copySurface."
> > >
> > > This reverts commit 6565506463.
> > >
> > > Reason for revert: seems to break things?
> > >
> > > Original change's description:
> > > > Remove support for copyAsDraw in gpu copySurface.
> > > >
> > > > The major changes on a higher lever are:
> > > > 1) The majority of all copies now go through GrSurfaceProxy::Copy which
> > > > takes in a proxy and returns a new one with the data copied to it. This
> > > > is the most common use case within Ganesh.
> > > >
> > > > 2) The backend copy calls no longer do draws, require origins to be the
> > > > same, and won't do any swizzling or adjustment of subrects. They are
> > > > all implemented to be dumb copy this data to this other spot.
> > > >
> > > > 3) The GrSurfaceContext copy call has now been moved to priv and renamed
> > > > copyNoDraw, and a new priv copyAsDraw was added to GrRenderTargetContext.
> > > >
> > > > 4) WritePixels and ReplaceRenderTarget both need to specifiy the destination
> > > > of copies. They are the only users (besides the GrSurfaceProxy::Copy) which
> > > > call the priv methods on GrSurfaceContext.
> > > >
> > > > Change-Id: Iaf1eb3a73ccaf39a75af77e281dae594f809186f
> > > > Reviewed-on: https://skia-review.googlesource.com/c/skia/+/217459
> > > > Reviewed-by: Brian Salomon <bsalomon@google.com>
> > > > Commit-Queue: Greg Daniel <egdaniel@google.com>
> > >
> > > TBR=egdaniel@google.com,bsalomon@google.com,robertphillips@google.com
> > >
> > > Change-Id: Id43aa8aa1451e794342e930441d9975b90e6b59f
> > > No-Presubmit: true
> > > No-Tree-Checks: true
> > > No-Try: true
> > > Reviewed-on: https://skia-review.googlesource.com/c/skia/+/218549
> > > 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: I1a96f85ae2ff7622a6b57406755d478e7fbcf56e
> > No-Presubmit: true
> > No-Tree-Checks: true
> > Reviewed-on: https://skia-review.googlesource.com/c/skia/+/218797
> > Reviewed-by: Brian Salomon <bsalomon@google.com>
> > Commit-Queue: Greg Daniel <egdaniel@google.com>
>
> TBR=egdaniel@google.com,bsalomon@google.com,robertphillips@google.com
>
> Change-Id: I310930a9df30535f45a065263a40239141e15562
> No-Presubmit: true
> No-Tree-Checks: true
> No-Try: true
> Reviewed-on: https://skia-review.googlesource.com/c/skia/+/219384
> 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: I88df4f19aa26ed77b5af4e25d138387cbabd1934
No-Presubmit: true
No-Tree-Checks: true
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/219386
Commit-Queue: Greg Daniel <egdaniel@google.com>
Reviewed-by: Greg Daniel <egdaniel@google.com>
This is to support the P016 and P010 YUV formats. Initially, only Vulkan and GL support these formats.
Change-Id: I4e896f1d3fb32207227a755517ae5a00a58e6045
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/219403
Reviewed-by: Greg Daniel <egdaniel@google.com>
Commit-Queue: Robert Phillips <robertphillips@google.com>
This reverts commit 93d0146dc6.
Reason for revert: revertting so we can revert previous change
Original change's description:
> Experimental: Add R_16 and RG_1616 to Ganesh
>
> This is to support the P016 and P010 YUV formats. Initially, only Vulkan and GL support these formats.
>
> Change-Id: Ie9609e59c12528079f8d379359ddb9bac85b6a29
> Reviewed-on: https://skia-review.googlesource.com/c/skia/+/218546
> Commit-Queue: Robert Phillips <robertphillips@google.com>
> Reviewed-by: Greg Daniel <egdaniel@google.com>
TBR=egdaniel@google.com,robertphillips@google.com
Change-Id: I00e1bf32404983b09f0933b2a9d65aa40a5ee754
No-Presubmit: true
No-Tree-Checks: true
No-Try: true
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/218959
Reviewed-by: Greg Daniel <egdaniel@google.com>
Commit-Queue: Greg Daniel <egdaniel@google.com>
This reverts commit 84ea04949c.
Reason for revert: nexus 7 and android one broken
Original change's description:
> Reland "Remove support for copyAsDraw in gpu copySurface."
>
> This reverts commit c5167c053b.
>
> Reason for revert: fixed
>
> Original change's description:
> > Revert "Remove support for copyAsDraw in gpu copySurface."
> >
> > This reverts commit 6565506463.
> >
> > Reason for revert: seems to break things?
> >
> > Original change's description:
> > > Remove support for copyAsDraw in gpu copySurface.
> > >
> > > The major changes on a higher lever are:
> > > 1) The majority of all copies now go through GrSurfaceProxy::Copy which
> > > takes in a proxy and returns a new one with the data copied to it. This
> > > is the most common use case within Ganesh.
> > >
> > > 2) The backend copy calls no longer do draws, require origins to be the
> > > same, and won't do any swizzling or adjustment of subrects. They are
> > > all implemented to be dumb copy this data to this other spot.
> > >
> > > 3) The GrSurfaceContext copy call has now been moved to priv and renamed
> > > copyNoDraw, and a new priv copyAsDraw was added to GrRenderTargetContext.
> > >
> > > 4) WritePixels and ReplaceRenderTarget both need to specifiy the destination
> > > of copies. They are the only users (besides the GrSurfaceProxy::Copy) which
> > > call the priv methods on GrSurfaceContext.
> > >
> > > Change-Id: Iaf1eb3a73ccaf39a75af77e281dae594f809186f
> > > Reviewed-on: https://skia-review.googlesource.com/c/skia/+/217459
> > > Reviewed-by: Brian Salomon <bsalomon@google.com>
> > > Commit-Queue: Greg Daniel <egdaniel@google.com>
> >
> > TBR=egdaniel@google.com,bsalomon@google.com,robertphillips@google.com
> >
> > Change-Id: Id43aa8aa1451e794342e930441d9975b90e6b59f
> > No-Presubmit: true
> > No-Tree-Checks: true
> > No-Try: true
> > Reviewed-on: https://skia-review.googlesource.com/c/skia/+/218549
> > 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: I1a96f85ae2ff7622a6b57406755d478e7fbcf56e
> No-Presubmit: true
> No-Tree-Checks: true
> Reviewed-on: https://skia-review.googlesource.com/c/skia/+/218797
> Reviewed-by: Brian Salomon <bsalomon@google.com>
> Commit-Queue: Greg Daniel <egdaniel@google.com>
TBR=egdaniel@google.com,bsalomon@google.com,robertphillips@google.com
Change-Id: I310930a9df30535f45a065263a40239141e15562
No-Presubmit: true
No-Tree-Checks: true
No-Try: true
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/219384
Reviewed-by: Greg Daniel <egdaniel@google.com>
Commit-Queue: Greg Daniel <egdaniel@google.com>
This reverts commit c5167c053b.
Reason for revert: fixed
Original change's description:
> Revert "Remove support for copyAsDraw in gpu copySurface."
>
> This reverts commit 6565506463.
>
> Reason for revert: seems to break things?
>
> Original change's description:
> > Remove support for copyAsDraw in gpu copySurface.
> >
> > The major changes on a higher lever are:
> > 1) The majority of all copies now go through GrSurfaceProxy::Copy which
> > takes in a proxy and returns a new one with the data copied to it. This
> > is the most common use case within Ganesh.
> >
> > 2) The backend copy calls no longer do draws, require origins to be the
> > same, and won't do any swizzling or adjustment of subrects. They are
> > all implemented to be dumb copy this data to this other spot.
> >
> > 3) The GrSurfaceContext copy call has now been moved to priv and renamed
> > copyNoDraw, and a new priv copyAsDraw was added to GrRenderTargetContext.
> >
> > 4) WritePixels and ReplaceRenderTarget both need to specifiy the destination
> > of copies. They are the only users (besides the GrSurfaceProxy::Copy) which
> > call the priv methods on GrSurfaceContext.
> >
> > Change-Id: Iaf1eb3a73ccaf39a75af77e281dae594f809186f
> > Reviewed-on: https://skia-review.googlesource.com/c/skia/+/217459
> > Reviewed-by: Brian Salomon <bsalomon@google.com>
> > Commit-Queue: Greg Daniel <egdaniel@google.com>
>
> TBR=egdaniel@google.com,bsalomon@google.com,robertphillips@google.com
>
> Change-Id: Id43aa8aa1451e794342e930441d9975b90e6b59f
> No-Presubmit: true
> No-Tree-Checks: true
> No-Try: true
> Reviewed-on: https://skia-review.googlesource.com/c/skia/+/218549
> 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: I1a96f85ae2ff7622a6b57406755d478e7fbcf56e
No-Presubmit: true
No-Tree-Checks: true
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/218797
Reviewed-by: Brian Salomon <bsalomon@google.com>
Commit-Queue: Greg Daniel <egdaniel@google.com>
This is to support the P016 and P010 YUV formats. Initially, only Vulkan and GL support these formats.
Change-Id: Ie9609e59c12528079f8d379359ddb9bac85b6a29
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/218546
Commit-Queue: Robert Phillips <robertphillips@google.com>
Reviewed-by: Greg Daniel <egdaniel@google.com>
This reverts commit 6565506463.
Reason for revert: seems to break things?
Original change's description:
> Remove support for copyAsDraw in gpu copySurface.
>
> The major changes on a higher lever are:
> 1) The majority of all copies now go through GrSurfaceProxy::Copy which
> takes in a proxy and returns a new one with the data copied to it. This
> is the most common use case within Ganesh.
>
> 2) The backend copy calls no longer do draws, require origins to be the
> same, and won't do any swizzling or adjustment of subrects. They are
> all implemented to be dumb copy this data to this other spot.
>
> 3) The GrSurfaceContext copy call has now been moved to priv and renamed
> copyNoDraw, and a new priv copyAsDraw was added to GrRenderTargetContext.
>
> 4) WritePixels and ReplaceRenderTarget both need to specifiy the destination
> of copies. They are the only users (besides the GrSurfaceProxy::Copy) which
> call the priv methods on GrSurfaceContext.
>
> Change-Id: Iaf1eb3a73ccaf39a75af77e281dae594f809186f
> Reviewed-on: https://skia-review.googlesource.com/c/skia/+/217459
> Reviewed-by: Brian Salomon <bsalomon@google.com>
> Commit-Queue: Greg Daniel <egdaniel@google.com>
TBR=egdaniel@google.com,bsalomon@google.com,robertphillips@google.com
Change-Id: Id43aa8aa1451e794342e930441d9975b90e6b59f
No-Presubmit: true
No-Tree-Checks: true
No-Try: true
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/218549
Reviewed-by: Greg Daniel <egdaniel@google.com>
Commit-Queue: Greg Daniel <egdaniel@google.com>
The major changes on a higher lever are:
1) The majority of all copies now go through GrSurfaceProxy::Copy which
takes in a proxy and returns a new one with the data copied to it. This
is the most common use case within Ganesh.
2) The backend copy calls no longer do draws, require origins to be the
same, and won't do any swizzling or adjustment of subrects. They are
all implemented to be dumb copy this data to this other spot.
3) The GrSurfaceContext copy call has now been moved to priv and renamed
copyNoDraw, and a new priv copyAsDraw was added to GrRenderTargetContext.
4) WritePixels and ReplaceRenderTarget both need to specifiy the destination
of copies. They are the only users (besides the GrSurfaceProxy::Copy) which
call the priv methods on GrSurfaceContext.
Change-Id: Iaf1eb3a73ccaf39a75af77e281dae594f809186f
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/217459
Reviewed-by: Brian Salomon <bsalomon@google.com>
Commit-Queue: Greg Daniel <egdaniel@google.com>
This is modeled on how we coordinate VkImage layout changes between
GrContext and clients.
A type GrGLTextureParameters is used to track the current parameter
state.
When a client creates a GrBackendTexture in order to wrap a resource
they created a new GrGLTextureParameters is created and the wrapped
GrGLTexture will share ownership.
When GrContext creates a non-wrapped GrGLTexture, the GrGLTexture
creates a new GrGLTextureParameters and any GrBackendTextures created
from that GrGLTexture will share ownership.
Clients indicate parameter changes by calling
GrBackendTexture::glTextureParametersModified().
We still assume all texture parameters may have changed after a call
to GrContext::resetContext() (for now). The "timestamp" that is used
to implement this has been moved from GrGpu to GrGLGpu as there were
no other use cases.
Change-Id: Ic24e00488fad254a29d5eec6890278b67df6efae
Bug: skia:7966
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/217385
Auto-Submit: Brian Salomon <bsalomon@google.com>
Commit-Queue: Greg Daniel <egdaniel@google.com>
Reviewed-by: Greg Daniel <egdaniel@google.com>
Also adds flexibility to unpremul the input, clamp the output, premul the
output or not.
Also fixes SkMatrix44 as a ctype.
The intent is to reuse this for rgb->yuv conversion in async rescale and
read.
Bug: skia:3962
Change-Id: I470d1cfebdbd79d8541b633c1747d510a5549ac4
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/217128
Reviewed-by: Brian Osman <brianosman@google.com>
Commit-Queue: Brian Salomon <bsalomon@google.com>
This reverts commit 186a295111.
Reason for revert: Metal bots failing
Original change's description:
> Change Metal to not take ownership of objects
>
> Prior to this change, Skia/Metal interfaces take ownership of the Metal
> objects passed in (that is, the caller should count passing the object
> to Skia as "freeing" the object).
>
> Change this behavior so that Skia/Metal retains its own separate
> ownership of the Metal objects.
>
> Make GrBackendTexture and GrBackendRenderTarget maintain their own
> references to the underlying MTLTexture by using the CFRetain/CFRelease
> interfaces. Do this by adding a private GrMtlBackendSurfaceInfo.
>
> Move GrMtlBackendSurfaceInfo (formerly GrMtlTextureInfo) out of the
> union in GrBackendTexture and GrBackendRenderTarget because unions
> cannot have nontrivial constructors and destructors (how fVkInfo isn't
> causing a compile error is unclear).
>
> Change-Id: Iae3719c0715825d86503d03c766e47f0f6015bdf
> Reviewed-on: https://skia-review.googlesource.com/c/skia/+/215685
> Commit-Queue: Jim Van Verth <jvanverth@google.com>
> Reviewed-by: Jim Van Verth <jvanverth@google.com>
TBR=egdaniel@google.com,jvanverth@google.com,bsalomon@google.com,ccameron@chromium.org
# Not skipping CQ checks because original CL landed > 1 day ago.
Change-Id: Ie569fe1938857706b5413876a9480ef1eb3314ea
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/216221
Reviewed-by: Jim Van Verth <jvanverth@google.com>
Commit-Queue: Jim Van Verth <jvanverth@google.com>
Prior to this change, Skia/Metal interfaces take ownership of the Metal
objects passed in (that is, the caller should count passing the object
to Skia as "freeing" the object).
Change this behavior so that Skia/Metal retains its own separate
ownership of the Metal objects.
Make GrBackendTexture and GrBackendRenderTarget maintain their own
references to the underlying MTLTexture by using the CFRetain/CFRelease
interfaces. Do this by adding a private GrMtlBackendSurfaceInfo.
Move GrMtlBackendSurfaceInfo (formerly GrMtlTextureInfo) out of the
union in GrBackendTexture and GrBackendRenderTarget because unions
cannot have nontrivial constructors and destructors (how fVkInfo isn't
causing a compile error is unclear).
Change-Id: Iae3719c0715825d86503d03c766e47f0f6015bdf
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/215685
Commit-Queue: Jim Van Verth <jvanverth@google.com>
Reviewed-by: Jim Van Verth <jvanverth@google.com>
This file pulls in Windows headers in a custom way, which is somewhat
awkward for a library header. The only use in include/ has been replaced
with a single forward declaration.
Change-Id: Ibef4cf7a2d1c9957a6a5b145b95aca1a6868cb5e
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/214689
Commit-Queue: Ben Wagner <bungeman@google.com>
Reviewed-by: Mike Klein <mtklein@google.com>
This partially undoes 774168efac "Allow
CCPR in DDL mode (take 2)". The issue appears to have been the use of a
not fully defined type (destructor could not be instantiated) because
the private fields were also dllexported, requiring the destructors to
exist in all translation units which included the header. Only the parts
of the class which are actually public are now marked as exported.
Change-Id: I8a79bd5d8962e94c24f7563d496744bb278153fa
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/214020
Commit-Queue: Ben Wagner <bungeman@google.com>
Reviewed-by: Mike Klein <mtklein@google.com>
Not used by anything else in the include directory, probably don't want
to accidentially expose it either.
Change-Id: I50d255e2cac43d8405305a825fd194bb36edd8fd
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/213826
Reviewed-by: Mike Klein <mtklein@google.com>
Commit-Queue: Ben Wagner <bungeman@google.com>
No public headers are using it anymore, so move it from include/private
into src/core where SkTSearch.cpp resides.
Change-Id: I4499c629487ff1b8c391b44708616d67567a3e9b
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/213674
Commit-Queue: Ben Wagner <bungeman@google.com>
Reviewed-by: Mike Klein <mtklein@google.com>
Introduce SkAutoMutexExclusive for SkMutex RAII.
Unsubclass SkMutex from SkBaseMutex to allow annotations
for class field mutexes separate from global mutexes.
Leave SkAutoMutexAcquire for handling global mutexes using
SkBaseMutex.
Test using GrSingleOwner.h.
Change-Id: I19d9d0ae0d05206cbb6ef137dc362969048c9c07
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/213136
Reviewed-by: Brian Osman <brianosman@google.com>
Commit-Queue: Herb Derby <herb@google.com>
Start out with spinlock. I tried to be more extensive, but some
of our abstractions confused the analysis. Will expand further
in following CLs.
Change-Id: I3e320c957d8ef427065a2b7e7d2187b7c6b0aef1
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/213060
Reviewed-by: Brian Osman <brianosman@google.com>
Commit-Queue: Herb Derby <herb@google.com>
Noticed that there was an include for <array> in there that didn't make
a lot of sense, so cleaned up the others which are hanging around from
older code which was in there.
Change-Id: I77acbb0914989e9bf67ab74dfd842a798ea592f3
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/206172
Reviewed-by: Herb Derby <herb@google.com>
Commit-Queue: Ben Wagner <bungeman@google.com>
Local Skia and DDL recording will always attempt to reduce opList splitting.
Android, Flutter, Google3 and non-DDL Chrome will not.
Note that this is a bit aggressive. Intermediate flushes based on memory usage have not yet been implemented.
The plan is to run this locally in Skia until the next Chrome branch and then enable it everywhere else (when intermediate flushes have been implemented).
OpList splitting reduction in Chrome is disabled in the following Chrome-side CL:
https://chromium-review.googlesource.com/c/chromium/src/+/1588756/ (Disable opList splitting reduction in Skia)
It is disabled in Android in:
https://googleplex-android-review.git.corp.google.com/c/platform/external/skia/+/7259923 (Update #defines to suppress Ganesh features in SkUserConfigManual.h)
It is disabled in Flutter and Google3 w/in this CL.
Change-Id: I59ff448d2c42629fab6cffccb2894d030c73431d
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/211101
Commit-Queue: Robert Phillips <robertphillips@google.com>
Reviewed-by: Greg Daniel <egdaniel@google.com>
This is a reland of 46d0f9aad1
Original change's description:
> ccpr: Set atlas proxy size to draw bounds rather than backing size
>
> Bug: skia:
> Change-Id: I6605754ecc5377b1c25847fdda478f8246979a2f
> Reviewed-on: https://skia-review.googlesource.com/c/skia/+/209808
> Commit-Queue: Chris Dalton <csmartdalton@google.com>
> Reviewed-by: Robert Phillips <robertphillips@google.com>
Bug: skia:
Change-Id: Ic13317fd021843961989a79050735c225702ad45
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/211181
Commit-Queue: Chris Dalton <csmartdalton@google.com>
Reviewed-by: Robert Phillips <robertphillips@google.com>
Add a shim to redirect until clients are updated
Change-Id: Ib43614e5620b1a24ca18187c1646a8ed1a9ee7a4
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/211003
Reviewed-by: Mike Klein <mtklein@google.com>
Commit-Queue: Brian Osman <brianosman@google.com>
Since explicit allocation is always enabled now, the resource allocator explicitly manages reuse of GrSurfaces and this flag isn't used/needed.
Change-Id: I5703bf4624e21f9aff9da76575f4ef757b1d2589
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/210140
Reviewed-by: Brian Salomon <bsalomon@google.com>
Commit-Queue: Robert Phillips <robertphillips@google.com>
Current strategy: everything from the top
Things to look at first are the manual changes:
- added tools/rewrite_includes.py
- removed -Idirectives from BUILD.gn
- various compile.sh simplifications
- tweak tools/embed_resources.py
- update gn/find_headers.py to write paths from the top
- update gn/gn_to_bp.py SkUserConfig.h layout
so that #include "include/config/SkUserConfig.h" always
gets the header we want.
No-Presubmit: true
Change-Id: I73a4b181654e0e38d229bc456c0d0854bae3363e
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/209706
Commit-Queue: Mike Klein <mtklein@google.com>
Reviewed-by: Hal Canary <halcanary@google.com>
Reviewed-by: Brian Osman <brianosman@google.com>
Reviewed-by: Florin Malita <fmalita@chromium.org>
Change-Id: I24fb1c38eb4cd9ec27b81a4e0010b1268442a29f
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/210063
Reviewed-by: Mike Klein <mtklein@google.com>
Commit-Queue: Hal Canary <halcanary@google.com>
Bug: skia:9012
Change-Id: I13ef8dea81a9d138b557be53b7adf19285fc9fce
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/209810
Commit-Queue: Mike Reed <reed@google.com>
Reviewed-by: Mike Klein <mtklein@google.com>
skcms is part of Skia's public API now. This attempts to recognize that,
and pave the way for moving the header to another location more easily
in a follow up CL, or - at a minimum - for clients that redistribute
Skia as a library + includes to relocate the skcms.h header as part of
that.
Change-Id: I15da63b0d4ab8916a71fb7e6ab3656db87252707
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/209640
Reviewed-by: Mike Klein <mtklein@google.com>
Reviewed-by: Kevin Lubick <kjlubick@google.com>
Commit-Queue: Brian Osman <brianosman@google.com>
This approach eliminates a lot of edge cases where ops (e.g., the SmallPathOp) are treating their proxies in a special manner (i.e., just holding a raw ref and never adding pendingIO). Given that the atlas managers are managing the lifetime of the proxies there is no reason for the GrResourceAllocator to be aware of them.
Pulled out of:
https://skia-review.googlesource.com/c/skia/+/208227 (Implement alternate method for determining recycle-ability of allocated GrSurfaces)
Change-Id: I0a8d3c2d56036d568b6d49ddc78c32761642d1c8
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/209419
Reviewed-by: Brian Salomon <bsalomon@google.com>
Commit-Queue: Robert Phillips <robertphillips@google.com>
This reverts commit 7bb47f2a2e.
Reason for revert: chrome roll maybe
Original change's description:
> Add explicit GrSurfaceProxy flag to skip explicit resource allocation
>
> This approach eliminates a lot of edge cases where ops (e.g., the SmallPathOp) are treating their proxies in a special manner (i.e., just holding a raw ref and never adding pendingIO). Given that the atlas managers are managing the lifetime of the proxies there is no reason for the GrResourceAllocator to be aware of them.
>
> Pulled out of:
>
> https://skia-review.googlesource.com/c/skia/+/208227 (Implement alternate method for determining recycle-ability of allocated GrSurfaces)
>
> Change-Id: Ia6bec5e8f5d5bc63e86ae011bcc3f8e061c066b2
> Reviewed-on: https://skia-review.googlesource.com/c/skia/+/209400
> Reviewed-by: Brian Salomon <bsalomon@google.com>
> Commit-Queue: Robert Phillips <robertphillips@google.com>
TBR=bsalomon@google.com,robertphillips@google.com
# Not skipping CQ checks because original CL landed > 1 day ago.
Change-Id: Id65bd176f56aa91ff76ec1979aef6206b7665d63
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/209600
Reviewed-by: Mike Klein <mtklein@google.com>
Commit-Queue: Mike Klein <mtklein@google.com>
This approach eliminates a lot of edge cases where ops (e.g., the SmallPathOp) are treating their proxies in a special manner (i.e., just holding a raw ref and never adding pendingIO). Given that the atlas managers are managing the lifetime of the proxies there is no reason for the GrResourceAllocator to be aware of them.
Pulled out of:
https://skia-review.googlesource.com/c/skia/+/208227 (Implement alternate method for determining recycle-ability of allocated GrSurfaces)
Change-Id: Ia6bec5e8f5d5bc63e86ae011bcc3f8e061c066b2
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/209400
Reviewed-by: Brian Salomon <bsalomon@google.com>
Commit-Queue: Robert Phillips <robertphillips@google.com>
This is a reland of e3b110dc6e
PS1 is the original, so best to diff against that.
This is the original with compiler workarounds.
Original change's description:
> align skvx::Vec<N,T> to N*sizeof(T)
>
> This increases the alignment of these vector types. I would have liked
> to keep the alignment minimal, but it's probably no big deal either way.
>
> In terms of code generation, it doesn't make much difference for x86 or
> ARMv8, but it seems hugely important for good ARMv7 NEON code. It's a
> ~10x difference for the bench I've been playing around with that spends
> most of its time in that SkOpts::blit_row_color32 routine.
>
> Bug: chromium:952502
> Change-Id: Ib12caad6b9b3f3f6e821ed70bfb57099db37b15f
> Reviewed-on: https://skia-review.googlesource.com/c/skia/+/208581
> Commit-Queue: Michael Ludwig <michaelludwig@google.com>
> Reviewed-by: Michael Ludwig <michaelludwig@google.com>
> Auto-Submit: Mike Klein <mtklein@google.com>
Bug: chromium:952502
Cq-Include-Trybots: skia.primary:Test-Win2016-MSVC-GCE-CPU-AVX2-x86-Release-All,Build-Debian9-GCC-mips64el-Debug
Change-Id: Ief99e14ab4de5a56840ed6bb326cf7669c51dc97
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/208681
Reviewed-by: Mike Klein <mtklein@google.com>
Commit-Queue: Mike Klein <mtklein@google.com>
Yet another surprising finding when looking at ARM code generation is
that passing these values to functions by const& does make a difference,
even when fully inlined. I can only guess that the compiler's somehow
more sure that way that the values won't change? Anyway, convert all
skvx functions that take Vec arguments to take const Vec& instead.
This tweak is enough to let the natural implementation of mull()
actually produce good code generation, so I've promoted that to SkVx.h
and added a unit test. Notice in the NEON case we've got a base case at
N=8 and two recursive cases, one down to 8 as usual when N > 8, but also
one up to 8 when N < 8.
This also is another big speedup for ARMv7 NEON, bringing it to nearly
the same speed as ARMv8 NEON on the same device.
Bug: chromium:952502
Change-Id: I0f19bab45cf02222ccc8090053ea2a4a380f1dfe
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/208582
Commit-Queue: Michael Ludwig <michaelludwig@google.com>
Auto-Submit: Mike Klein <mtklein@google.com>
Reviewed-by: Michael Ludwig <michaelludwig@google.com>
This reverts commit e3b110dc6e.
Reason for revert: bot failures
Original change's description:
> align skvx::Vec<N,T> to N*sizeof(T)
>
> This increases the alignment of these vector types. I would have liked
> to keep the alignment minimal, but it's probably no big deal either way.
>
> In terms of code generation, it doesn't make much difference for x86 or
> ARMv8, but it seems hugely important for good ARMv7 NEON code. It's a
> ~10x difference for the bench I've been playing around with that spends
> most of its time in that SkOpts::blit_row_color32 routine.
>
> Bug: chromium:952502
> Change-Id: Ib12caad6b9b3f3f6e821ed70bfb57099db37b15f
> Reviewed-on: https://skia-review.googlesource.com/c/skia/+/208581
> Commit-Queue: Michael Ludwig <michaelludwig@google.com>
> Reviewed-by: Michael Ludwig <michaelludwig@google.com>
> Auto-Submit: Mike Klein <mtklein@google.com>
TBR=mtklein@google.com,michaelludwig@google.com
Change-Id: I72357b9775685efcc2cd75db220711c8145b8ac4
No-Presubmit: true
No-Tree-Checks: true
No-Try: true
Bug: chromium:952502
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/208680
Reviewed-by: Mike Klein <mtklein@google.com>
Commit-Queue: Mike Klein <mtklein@google.com>
This increases the alignment of these vector types. I would have liked
to keep the alignment minimal, but it's probably no big deal either way.
In terms of code generation, it doesn't make much difference for x86 or
ARMv8, but it seems hugely important for good ARMv7 NEON code. It's a
~10x difference for the bench I've been playing around with that spends
most of its time in that SkOpts::blit_row_color32 routine.
Bug: chromium:952502
Change-Id: Ib12caad6b9b3f3f6e821ed70bfb57099db37b15f
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/208581
Commit-Queue: Michael Ludwig <michaelludwig@google.com>
Reviewed-by: Michael Ludwig <michaelludwig@google.com>
Auto-Submit: Mike Klein <mtklein@google.com>
This is preparing the ground for having proxies be atomically reffed
Change-Id: Ice48c41120d4ec10a758f696a69028b6363e8a7e
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/207762
Reviewed-by: Greg Daniel <egdaniel@google.com>
Commit-Queue: Robert Phillips <robertphillips@google.com>
Try: out/release/nanobench --match _charToGlyph
Pseudo plan to use this:
- attach to whatever typeface backends need it (probably just freetype)
- have a purge/limiting scheme (e.g. only cache N entries)
- if we care, make the search fancier (e.g. binary, slope, etc.)
Bug: 951647
Change-Id: Ib1042ca5891d2742499faf1314579c402121a855
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/207703
Reviewed-by: Herb Derby <herb@google.com>
Commit-Queue: Mike Reed <reed@google.com>
These replicate the base logic of Sk4px::Wide::div255() and
Sk4px::approxMulDiv255(), and will come in handy replacing them.
No platform specializations yet... want to remind myself what
codegen they get from these vanilla versions first, and then
I'll fill in the platform specific stuff as needed. The tests
should cover everything pretty exhaustively.
Change-Id: I5854d1bc0902a85cbb2351f669c4da7cc31a8775
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/207683
Commit-Queue: Mike Klein <mtklein@google.com>
Commit-Queue: Michael Ludwig <michaelludwig@google.com>
Reviewed-by: Michael Ludwig <michaelludwig@google.com>
Auto-Submit: Mike Klein <mtklein@google.com>
For some reason, Clang can infer <N,T> but GCC can't.
No big deal... we know exactly the ones we want anyway.
Change-Id: I15ba4d4edbd3bc0f37ebe3c2b6e411726cd9fb69
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/207341
Commit-Queue: Mike Klein <mtklein@google.com>
Reviewed-by: Michael Ludwig <michaelludwig@google.com>
Was starting to use this and ran into a few problems with clashing
symbols, namely SI and cast(). Seemed simple enough to not use SI,
and to move all the free-standing types into skvx: skvx::cast,
skvx::shuffle, etc.
Change-Id: Ia5d8ef6d0ae5375bf80d76be88d16f0c9cde56e7
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/207340
Commit-Queue: Mike Klein <mtklein@google.com>
Auto-Submit: Mike Klein <mtklein@google.com>
Reviewed-by: Michael Ludwig <michaelludwig@google.com>
This is rather ham-fisted but I would like to have a short-term fix for the "always-flush" perf regressions.
Change-Id: I508e2d725ac18f50318baf31bc3243fe932c724b
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/206697
Reviewed-by: Greg Daniel <egdaniel@google.com>
Commit-Queue: Robert Phillips <robertphillips@google.com>
