VFPv4 gives us two interesting features:
- FMA
- f16<->f32 conversions
Even without FMAs, NEON still has non-fused MLA instructions. We don't
really care about the fusedness of those mul-adds, so losing FMA here is
kind of no big deal.
We already maintain portable code to do f16<->f32 conversions, so it's
not much of a maintanence hit to use that instead of the native
instructions. To my knowledge software F16 rendering is not a
performance critical mode of operation for any of our users.
This drops our minimum requirement to basically just having NEON.
Devices like the Nexus 7 2012 will now take SkJumper fast paths
instead of portable code. (Though actually, we've only ever
required NEON for _lowp... only the float code also needed vfpv4).
The main file to look at here is actually SkJumper_vectors.h,
where you will see all the substantive changes. The rest just
kind of tears down most of the old complexity, add adds ABI
to put just a little of it back. :)
Change-Id: Ia9237117698729c91e5fa51126baf80748093bf4
Bug: skia:
Reviewed-on: https://skia-review.googlesource.com/83521
Commit-Queue: Mike Klein <mtklein@chromium.org>
Reviewed-by: Florin Malita <fmalita@chromium.org>
c7abc08034..629a64498d
$ git log c7abc0803..629a64498 --date=short --no-merges --format='%ad %ae %s'
2017-12-11 oetuaho Fix HLSL integer pow workaround
Created with:
roll-dep skia/third_party/externals/angle2
BUG=793115
The AutoRoll server is located here: https://angle-skia-roll.skia.org
Documentation for the AutoRoller is here:
https://skia.googlesource.com/buildbot/+/master/autoroll/README.md
If the roll is causing failures, please contact the current sheriff, who should
be CC'd on the roll, and stop the roller if necessary.
CQ_INCLUDE_TRYBOTS=skia.primary:Perf-Win10-Clang-AlphaR2-GPU-RadeonR9M470X-x86_64-Debug-All-ANGLE,Perf-Win10-Clang-NUC5i7RYH-GPU-IntelIris6100-x86_64-Debug-All-ANGLE,Perf-Win10-Clang-NUC6i5SYK-GPU-IntelIris540-x86_64-Debug-All-ANGLE,Perf-Win10-Clang-NUCD34010WYKH-GPU-IntelHD4400-x86_64-Debug-All-ANGLE,Perf-Win10-Clang-ShuttleC-GPU-GTX960-x86_64-Debug-All-ANGLE,Test-Win10-Clang-AlphaR2-GPU-RadeonR9M470X-x86_64-Debug-All-ANGLE,Test-Win10-Clang-NUC6i5SYK-GPU-IntelIris540-x86_64-Debug-All-ANGLE,Test-Win10-Clang-NUCD34010WYKH-GPU-IntelHD4400-x86_64-Debug-All-ANGLE,Test-Win10-Clang-ShuttleC-GPU-GTX960-x86_64-Debug-All-ANGLE,Build-Debian9-Clang-x86_64-Release-ANGLE
TBR=liyuqian@google.com
Change-Id: I73a2bee4cae25f79d0ccd1fe68e13bc62f9f7a63
Reviewed-on: https://skia-review.googlesource.com/83800
Reviewed-by: angle-deps-roller . <angle-deps-roller@chromium.org>
Commit-Queue: angle-deps-roller . <angle-deps-roller@chromium.org>
To enable, set skia_embed_resources=true in args.gn.
Also add *-EmbededResouces bots.
Change-Id: Ia69b26e926a3ad4676a4fa021894432ea2104538
Reviewed-on: https://skia-review.googlesource.com/82626
Commit-Queue: Hal Canary <halcanary@google.com>
Reviewed-by: Mike Klein <mtklein@google.com>
Add GrGLAssembleInterface with legacy bare pointer return.
This allows existing clients of GrGLAssembleInterface to roll Skia without
code changes.
Change-Id: I0764a9f4583e554fff5574889adcc6fe004db159
Reviewed-on: https://skia-review.googlesource.com/83564
Commit-Queue: Brian Salomon <bsalomon@google.com>
Reviewed-by: Brian Salomon <bsalomon@google.com>
Add more examples and docs for SkImage; still a ways to go.
Fix bit-rotted examples.
Add typedef support.
Add json driver to pick files to work on; remove special-casing.
Fix unordered map traversal that made md output unreliable.
TBR=rmistry@google.com
Docs-Preview: https://skia.org/?cl=80060
Bug: skia:6898
Change-Id: Ib8eb9fdfa5a9db61c8332e657fa2e2f4b96a665f
Reviewed-on: https://skia-review.googlesource.com/80060
Reviewed-by: Cary Clark <caryclark@skia.org>
Commit-Queue: Cary Clark <caryclark@skia.org>
9b88991c72..c7abc08034
$ git log 9b88991c7..c7abc0803 --date=short --no-merges --format='%ad %ae %s'
2017-07-04 cwallez Implement EGL_ANGLE_iosurface_client_buffer
Created with:
roll-dep skia/third_party/externals/angle2
The AutoRoll server is located here: https://angle-skia-roll.skia.org
Documentation for the AutoRoller is here:
https://skia.googlesource.com/buildbot/+/master/autoroll/README.md
If the roll is causing failures, please contact the current sheriff, who should
be CC'd on the roll, and stop the roller if necessary.
CQ_INCLUDE_TRYBOTS=skia.primary:Perf-Win10-Clang-AlphaR2-GPU-RadeonR9M470X-x86_64-Debug-All-ANGLE,Perf-Win10-Clang-NUC5i7RYH-GPU-IntelIris6100-x86_64-Debug-All-ANGLE,Perf-Win10-Clang-NUC6i5SYK-GPU-IntelIris540-x86_64-Debug-All-ANGLE,Perf-Win10-Clang-NUCD34010WYKH-GPU-IntelHD4400-x86_64-Debug-All-ANGLE,Perf-Win10-Clang-ShuttleC-GPU-GTX960-x86_64-Debug-All-ANGLE,Test-Win10-Clang-AlphaR2-GPU-RadeonR9M470X-x86_64-Debug-All-ANGLE,Test-Win10-Clang-NUC6i5SYK-GPU-IntelIris540-x86_64-Debug-All-ANGLE,Test-Win10-Clang-NUCD34010WYKH-GPU-IntelHD4400-x86_64-Debug-All-ANGLE,Test-Win10-Clang-ShuttleC-GPU-GTX960-x86_64-Debug-All-ANGLE,Build-Debian9-Clang-x86_64-Release-ANGLE
TBR=liyuqian@google.com
Change-Id: I30c6dc83cb9a8a90514e99b32ca4a530ec0e2176
Reviewed-on: https://skia-review.googlesource.com/83505
Reviewed-by: angle-deps-roller . <angle-deps-roller@chromium.org>
Commit-Queue: angle-deps-roller . <angle-deps-roller@chromium.org>
This avoids bugs when other directories (like the Driver SDK) show up in
those folders.
Bug: skia:7395
Change-Id: Iee316a7daf8d71223b999de736d63e1dc7fa31f7
Reviewed-on: https://skia-review.googlesource.com/83542
Commit-Queue: Brian Osman <brianosman@google.com>
Reviewed-by: Mike Klein <mtklein@chromium.org>
Here's the tiny performance gain:
$python tools/calmbench/calmbench.py firstsecond --extraarg "-m conic"
firstsecond (compared to master) is likely
4.23% faster in gradient_conicalOut_clamp_3color
4.23% faster in gradient_conicalOutZero_clamp_3color
4.79% faster in gradient_conical_clamp_shallow_dither
6.04% faster in gradient_conical_clamp_3color
6.04% faster in gradient_conicalZero_clamp_3color
6.42% faster in gradient_conicalOut_clamp
6.43% faster in gradient_conicalOutZero_clamp
6.74% faster in gradient_conical_clamp
6.98% faster in gradient_conical_clamp_shallow
6.98% faster in gradient_conicalZero_clamp
Bug: skia:
Change-Id: Id74866908b99753ed8b16a657d3f67c9255d0043
Reviewed-on: https://skia-review.googlesource.com/76561
Commit-Queue: Yuqian Li <liyuqian@google.com>
Reviewed-by: Mike Klein <mtklein@chromium.org>
Reviewed-by: Florin Malita <fmalita@chromium.org>
Bug:793603
Change-Id: I040d775fcf76da095ea2a25c33408508b3466fd7
Reviewed-on: https://skia-review.googlesource.com/83280
Commit-Queue: Mike Reed <reed@google.com>
Reviewed-by: Mike Klein <mtklein@chromium.org>
This patch is to turn on PGO for skia on Android, which will provide a
performance improvement of 7.6% for hwui when applied PGO for both.
The patch specified a skia.profdata file, which locates in internal
google_data/pgo_profile directory, to work as the profile to feed PGO.
This profdata can be re-collected with PGO build system support.
PGO can be turned off by setting ANDROID_PGO_NO_PROFILE_USE environment
variable or set variable to false.
Test: Build skia successfully and verified the performance improvement
on device through benchmark.
BUG=skia:
Change-Id: I826f417569b2853630f6d4fcce236b5bc36547fa
Reviewed-on: https://skia-review.googlesource.com/82880
Commit-Queue: Derek Sollenberger <djsollen@google.com>
Reviewed-by: Derek Sollenberger <djsollen@google.com>
Reviewed-by: John Reck <jreck@google.com>
Add a flag that hints, which lattice rectangles are solid colors.
Draw solid rectangles and 1x1 rectangles with drawRect.
Test: Measured performance of a ninepatch drawn by HWUI
Bug: b/69796044
Change-Id: Ib3b00ca608da42fa9f2d2038cc126a978421ec7c
Reviewed-on: https://skia-review.googlesource.com/79821
Commit-Queue: Stan Iliev <stani@google.com>
Reviewed-by: Derek Sollenberger <djsollen@google.com>
The Nexus5x took these jobs in https://skia-review.googlesource.com/c/skia/+/82624
Also filed skia:7394 for missing coverage of "--gpuThreads 0".
Bug: skia:7382
Change-Id: If0fd823722d8b8cc3845f8d96a0a285fec935f74
No-Try: true
Reviewed-on: https://skia-review.googlesource.com/83360
Reviewed-by: Kevin Lubick <kjlubick@google.com>
Commit-Queue: Ben Wagner <benjaminwagner@google.com>
Removes the concept of a configurable "default" interface and makes the default
always be the "native" interface.
Also removes unused functions: GrGLInterfaceAddTestDebugMarker and
GrGLInterface::NewClone.
Keeps around legacy GrGLCreateNativeInterface() until clients can be weened.
Change-Id: I4a3bdafa8cf8c68ed13318393abd55686b045ccb
Reviewed-on: https://skia-review.googlesource.com/83000
Reviewed-by: Robert Phillips <robertphillips@google.com>
Commit-Queue: Brian Salomon <bsalomon@google.com>
Viewer still has plenty of code that uses ImGui to create application
specific UI, but the structural code that forwards input to ImGui, and
converts per-frame ImGui rendering data to Skia draw commands is now in
a single component that can be reused in any sk_app-based application.
Bug: skia:
Change-Id: Ic14ece659d4af8ee13b69c638bdaf7df6c24f5c0
Reviewed-on: https://skia-review.googlesource.com/82627
Commit-Queue: Brian Osman <brianosman@google.com>
Reviewed-by: Jim Van Verth <jvanverth@google.com>
It isn't necessary for the vertex shader to know all the points
because everything happens in the geometry shader. It's simpler to use
regular vertex arrays instead.
Bug: skia:
Change-Id: I7bf83180476fbe0ab01492611cd72e72b3f7d4f2
Reviewed-on: https://skia-review.googlesource.com/82881
Reviewed-by: Brian Salomon <bsalomon@google.com>
Commit-Queue: Chris Dalton <csmartdalton@google.com>
This also makes it possible to manage other parts of viewer, etc (like
the stats screen, command set, even samples) as additional layers in the
stack. For now, it just removes a lot of boilerplate.
Bug: skia:
Change-Id: Ic2f80690fc76c683b3736287dc2b738c50d38614
Reviewed-on: https://skia-review.googlesource.com/82688
Reviewed-by: Jim Van Verth <jvanverth@google.com>
Commit-Queue: Brian Osman <brianosman@google.com>
So there's no need to include the image name or path in the GM name.
Change-Id: Idc510906ae77c5c73756f41bc13e859e231e9139
Reviewed-on: https://skia-review.googlesource.com/83320
Reviewed-by: Hal Canary <halcanary@google.com>
Commit-Queue: Mike Klein <mtklein@chromium.org>
This reverts commit a7fa3377d2.
Reason for revert: lots of crashing GPU bots.
Original change's description:
> add experimental bilerp_clamp_8888 stage
>
> It looks like we can specialize hot image shaders into their
> own single stages for a good speedup on both x86 and ARM.
>
> I've started here with bilerp_clamp_8888, and will
> follow up with bgra and 565, and lowp versions of those,
> and probably also the same for nearest neighbors.
>
> All pixels are identical in GMs.
>
> Change-Id: I2f6995767cd38053d670b8d0bfdb71b687803d70
> Reviewed-on: https://skia-review.googlesource.com/82100
> Reviewed-by: Yuqian Li <liyuqian@google.com>
> Commit-Queue: Mike Klein <mtklein@chromium.org>
TBR=mtklein@chromium.org,mtklein@google.com,liyuqian@google.com
Change-Id: If70abb91b69bcd781e395dd3ac05ff1eebb1169f
No-Presubmit: true
No-Tree-Checks: true
No-Try: true
Reviewed-on: https://skia-review.googlesource.com/83340
Reviewed-by: Mike Klein <mtklein@chromium.org>
Commit-Queue: Mike Klein <mtklein@chromium.org>
No-Try: true
Change-Id: I9c07250623603be1baccdc177191326017784aa2
Reviewed-on: https://skia-review.googlesource.com/82604
Reviewed-by: Eric Boren <borenet@google.com>
Reviewed-by: Yuqian Li <liyuqian@google.com>
Commit-Queue: Ben Wagner <benjaminwagner@google.com>
It looks like we can specialize hot image shaders into their
own single stages for a good speedup on both x86 and ARM.
I've started here with bilerp_clamp_8888, and will
follow up with bgra and 565, and lowp versions of those,
and probably also the same for nearest neighbors.
All pixels are identical in GMs.
Change-Id: I2f6995767cd38053d670b8d0bfdb71b687803d70
Reviewed-on: https://skia-review.googlesource.com/82100
Reviewed-by: Yuqian Li <liyuqian@google.com>
Commit-Queue: Mike Klein <mtklein@chromium.org>
-1 is not a valid value for the GrTextureDomain::Mode enum, so certain
versions of Clang warn that the comparison is a no-op:
../../third_party/skia/src/gpu/effects/GrTextureDomain.cpp:70:23: warning:
comparison of constant -1 with expression of type 'GrTextureDomain::Mode' is
always false [-Wtautological-constant-out-of-range-compare]
SkASSERT((Mode)-1 == fMode || textureDomain.mode() == fMode);
~~~~~~~~ ^ ~~~~~
Bug: chromium:793189
Change-Id: Iba49ae15622285a62ad218dbdd78b0846338b7cb
Reviewed-on: https://skia-review.googlesource.com/82962
Reviewed-by: Chris Dalton <csmartdalton@google.com>
Commit-Queue: Chris Dalton <csmartdalton@google.com>
-1 is not a valid value for the GrSurfaceOrigin enum, and certain
versions of Clang warn that the comparison is effectively a no-op:
../../third_party/skia/src/gpu/GrSurfaceProxy.cpp:547:38: warning: comparison
of constant -1 with expression of type 'GrSurfaceOrigin' is always true
[-Wtautological-constant-out-of-range-compare]
SkASSERT(kGrUnknownSurfaceOrigin != fProxy->origin());
~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~~~~~~~~
Instead of trying to force -1 in there, drop the assert and initialize with
top-left as origin.
Bug: chromium:793189
Change-Id: I4cb6720d567f6c5650a19df33d3c77f2d738a516
Reviewed-on: https://skia-review.googlesource.com/82961
Reviewed-by: Chris Dalton <csmartdalton@google.com>
Commit-Queue: Chris Dalton <csmartdalton@google.com>
This reverts commit cca2300559.
Reason for revert: think I guessed wrong about g32 -- unreverting
Original change's description:
> Revert "resources: remove most uses of GetResourcePath()"
>
> This reverts commit 5093a539de.
>
> Reason for revert: google3 seems broken
>
> Original change's description:
> > resources: remove most uses of GetResourcePath()
> >
> > Going forward, we will standardize on GetResourceAsData(), which will
> > make it easier to run tests in environments without access to the
> > filesystem.
> >
> > Also: GetResourceAsData() complains when a resource is missing.
> > This is usually an error.
> >
> > Change-Id: Iaf70b71b0ca5ed8cd1a5538a60ef185ae8736188
> > Reviewed-on: https://skia-review.googlesource.com/82642
> > Reviewed-by: Hal Canary <halcanary@google.com>
> > Commit-Queue: Hal Canary <halcanary@google.com>
>
> TBR=halcanary@google.com,scroggo@google.com
>
> Change-Id: Ic5a7c0167c995a672e6b06dc92abe00564432214
> No-Presubmit: true
> No-Tree-Checks: true
> No-Try: true
> Reviewed-on: https://skia-review.googlesource.com/83001
> Reviewed-by: Mike Reed <reed@google.com>
> Commit-Queue: Mike Reed <reed@google.com>
TBR=halcanary@google.com,scroggo@google.com,reed@google.com
Change-Id: I5a46e4de61186a8a5eb9cacd3275e24e311d5a07
No-Presubmit: true
No-Tree-Checks: true
No-Try: true
Reviewed-on: https://skia-review.googlesource.com/82942
Reviewed-by: Mike Reed <reed@google.com>
Commit-Queue: Mike Reed <reed@google.com>
58ba6bf5ec..9b88991c72
$ git log 58ba6bf5e..9b88991c7 --date=short --no-merges --format='%ad %ae %s'
2017-12-08 kainino Disable DIRTY_BIT_PROGRAM_UNIFORM_BUFFERS
Created with:
roll-dep skia/third_party/externals/angle2
The AutoRoll server is located here: https://angle-skia-roll.skia.org
Documentation for the AutoRoller is here:
https://skia.googlesource.com/buildbot/+/master/autoroll/README.md
If the roll is causing failures, please contact the current sheriff, who should
be CC'd on the roll, and stop the roller if necessary.
CQ_INCLUDE_TRYBOTS=skia.primary:Perf-Win10-Clang-AlphaR2-GPU-RadeonR9M470X-x86_64-Debug-All-ANGLE,Perf-Win10-Clang-NUC5i7RYH-GPU-IntelIris6100-x86_64-Debug-All-ANGLE,Perf-Win10-Clang-NUC6i5SYK-GPU-IntelIris540-x86_64-Debug-All-ANGLE,Perf-Win10-Clang-NUCD34010WYKH-GPU-IntelHD4400-x86_64-Debug-All-ANGLE,Perf-Win10-Clang-ShuttleC-GPU-GTX960-x86_64-Debug-All-ANGLE,Test-Win10-Clang-AlphaR2-GPU-RadeonR9M470X-x86_64-Debug-All-ANGLE,Test-Win10-Clang-NUC6i5SYK-GPU-IntelIris540-x86_64-Debug-All-ANGLE,Test-Win10-Clang-NUCD34010WYKH-GPU-IntelHD4400-x86_64-Debug-All-ANGLE,Test-Win10-Clang-ShuttleC-GPU-GTX960-x86_64-Debug-All-ANGLE,Build-Debian9-Clang-x86_64-Release-ANGLE
TBR=rmistry@google.com
Change-Id: I895da35eafadb08a13c873f0542c3ee3b1f83fa4
Reviewed-on: https://skia-review.googlesource.com/83041
Reviewed-by: angle-deps-roller . <angle-deps-roller@chromium.org>
Commit-Queue: angle-deps-roller . <angle-deps-roller@chromium.org>
The new bot:
1) Extracts all fiddles out of markdown files.
2) Forces fiddle.skia.org to compile all those fiddles and get output in JSON.
3) Scans the output and reports any compiletime/runtime errors.
4) Updates markdown in site/user/api/ using the new hashes (if any) from fiddle.skia.org.
To ensure the bot does not check in unexpected changes:
* upload_md.py makes sure that all modified files are under site/user/api/. It errors out if this is not the case.
* CLs with docs only changes normally have 'NoTry: true' added to their descriptions. This is not done for the new bot because we want to run it through trybots to make sure nothing unexpected snuck in.
* rmistry@ and caryclark@ are automatically CC'ed on all uploaded changes. I plan to watch the bot closely for at least a couple of weeks.
Bug: skia:7310
Change-Id: I759fc8bae1e32a6f175e3b1a895947d14ca5fe3a
Reviewed-on: https://skia-review.googlesource.com/79941
Commit-Queue: Ravi Mistry <rmistry@google.com>
Reviewed-by: Eric Boren <borenet@google.com>
This reverts commit 5093a539de.
Reason for revert: google3 seems broken
Original change's description:
> resources: remove most uses of GetResourcePath()
>
> Going forward, we will standardize on GetResourceAsData(), which will
> make it easier to run tests in environments without access to the
> filesystem.
>
> Also: GetResourceAsData() complains when a resource is missing.
> This is usually an error.
>
> Change-Id: Iaf70b71b0ca5ed8cd1a5538a60ef185ae8736188
> Reviewed-on: https://skia-review.googlesource.com/82642
> Reviewed-by: Hal Canary <halcanary@google.com>
> Commit-Queue: Hal Canary <halcanary@google.com>
TBR=halcanary@google.com,scroggo@google.com
Change-Id: Ic5a7c0167c995a672e6b06dc92abe00564432214
No-Presubmit: true
No-Tree-Checks: true
No-Try: true
Reviewed-on: https://skia-review.googlesource.com/83001
Reviewed-by: Mike Reed <reed@google.com>
Commit-Queue: Mike Reed <reed@google.com>
Using full paths in skia.h causes "file not found" with Goma. All other
Builds seem fine without the path, so I changed find_headers.py to use
the basename.
Change-Id: Ib520e91a92ebffe36a736eb53f643d359f5bb2ce
Reviewed-on: https://skia-review.googlesource.com/79360
Reviewed-by: Greg Daniel <egdaniel@google.com>
Reviewed-by: Eric Boren <borenet@google.com>
Commit-Queue: Ben Wagner <benjaminwagner@google.com>
Going forward, we will standardize on GetResourceAsData(), which will
make it easier to run tests in environments without access to the
filesystem.
Also: GetResourceAsData() complains when a resource is missing.
This is usually an error.
Change-Id: Iaf70b71b0ca5ed8cd1a5538a60ef185ae8736188
Reviewed-on: https://skia-review.googlesource.com/82642
Reviewed-by: Hal Canary <halcanary@google.com>
Commit-Queue: Hal Canary <halcanary@google.com>
This reverts commit 1a104bce20.
Change (from first version) is
- only signal error in readbuffer for corrupt stream, not default fonts
- change test to ensure a non-null typeface (i.e. MakeDefault())
Bug: skia:
Change-Id: I325445b56b0a402e1b89a2439df06e92314c793f
Reviewed-on: https://skia-review.googlesource.com/82687
Reviewed-by: Mike Reed <reed@google.com>
Commit-Queue: Mike Reed <reed@google.com>
