Need to land SK_SUPPORT_LEGACY_SCALAR_MAPPOINTS in chrome to suppress Affine
version which causes slight differences (which will need to be rebaselined)
BUG=skia:
Review URL: https://codereview.chromium.org/1045493002
Currently, the GPU-side image filter implementation creates exact-match
textures for the offscreen backing stores for saveLayer(). This is
because several filters have GPU implementations which depend on the
texture coordinates being 0..1.
The fix is three-fold:
1) Store the actual requested size in the SkGpuDevice, so that when
wrapping it in an SkBitmap for passing to filterImage(), we can give
it the original size.
2) Fix the filters (SkMagnifierImageFilter, more TBD) whose GPU
implementation depends on 0..1 texture coordinates.
3) Remove the exception for GPU-side image filters in
SkCanvas::internalSaveLayer().
N.B.: this change will cause some minor pixel diffs in the
GPU results of the following GMs (and possibly more):
matriximagefilter, matrixconvolution, imagefiltersscaled,
lighting, imagemagnifier, filterfastbounds,
complexclip_aa_Layer_invert, complexclip_aa_layer,
complexclip_bw_layer_invert, complexclip_bw_layer.
BUG=skia:3532
Review URL: https://codereview.chromium.org/1034733002
I'd like to add a new API to SkStream for peeking - i.e. reading some
bytes without advancing the stream. This will be implemented for the
streams where it makes sense. I think the function should look
something like the following:
size_t peek(void* buffer, size_t bytesToRead) {
return this->onPeek(buffer, bytesToRead);
}
virtual size_t onPeek(void* buffer, size_t bytesToRead) {
return 0; // unimplemented base class.
}
In order to avoid confusion, I'd like to remove SkMemoryStream::peek(),
which is not currently used internally, by Chrome, or by Android as far
as I can tell. There is also another function does the same thing:
getPosition().
BUG=skia:3257
Review URL: https://codereview.chromium.org/1039373002
Duplicate code from the HWUI backends for DM and nanobench
moves into a single place, saving a hundred lines or more of
cut-and-paste.
There's some indication that this increases the incidence of
SkCanvas "Unable to find device for layer." warnings, but no
clear degradation in test results.
R=djsollen@google.com,mtklein@google.com
BUG=skia:3589
Review URL: https://codereview.chromium.org/1036303002
In some cases, resolving an intersection can cause a vertex to go
slightly out-of-order with edges which have already been processed.
This doesn't cause any algorithmic errors, but it's difficult to detect
without impacting performance significantly.
Also, the GPU infrastructure fires asserts when attempting
to allocate 0-length vertex buffers. Early-out instead,
since there's nothing to draw.
Review URL: https://codereview.chromium.org/1032253005
Add and test trunc(), which is what get() used to be before rounding.
Using trunc() is a ~40% speedup on our linear gradient bench.
#neon #floats
BUG=skia:3592
#n5
#n9
CQ_INCLUDE_TRYBOTS=client.skia.android:Test-Android-Nexus5-Adreno330-Arm7-Debug-Trybot;client.skia.android:Test-Android-Nexus9-TegraK1-Arm64-Release-Trybot
Review URL: https://codereview.chromium.org/1032243002
Displaying the offset into an SkPicture hasn't worked for a while so this CL deletes the feature.
When "Save Layer" was renamed to "SaveLayer" the code that computes the indent in the list view was broken. This CL patches the problem.
Review URL: https://codereview.chromium.org/1034733004
Replace the implicit curve intersection with a geometric curve intersection. The implicit intersection proved mathematically unstable and took a long time to zero in on an answer.
Use pointers instead of indices to refer to parts of curves. Indices required awkward renumbering.
Unify t and point values so that small intervals can be eliminated in one pass.
Break cubics up front to eliminate loops and cusps.
Make the Simplify and Op code more regular and eliminate arbitrary differences.
Add a builder that takes an array of paths and operators.
Delete unused code.
BUG=skia:3588
R=reed@google.com
Review URL: https://codereview.chromium.org/1037573004
Introduce a paint filter proxy base class as a SkDrawFilter replacement,
and convert SkDebugCanvas to use the new approach.
BUG=skia:3587
R=reed@google.com,mtklein@google.com,robertphillips@google.com,tomhudson@google.com
Review URL: https://codereview.chromium.org/1032173002
It can be silenced or not with one flag.
Always print when building for the android framework.
Also remove the meaningless define of override to override.
BUG=skia:3257
Review URL: https://codereview.chromium.org/1032093004
./gyp_skia -Dskia_fast=1 will
- always optimize for the current machine as much as possible
- drop the frame pointer
- optimize floating point arithmetic ignoring IEEE compliance
This allows things like 3-argument VEX prefix SSE instructions and
NEON autovectorization, and can give some seriously helpful clues
about how to rearrange floating point math for speed.
I've been having trouble with LTO linking, so I'll leave that out for now.
I don't think we should set up bots with this mode. -ffast-math means
I'm forced to run DM with --match ~Blend ~Path ~Rect ~Math ~Scalar
~Matrix ~Point before it passes, which is a disconcerting chunk of tests
to disable (including all of PathOps).
BUG=skia:
Review URL: https://codereview.chromium.org/1036533003
Am I going nuts or can we get this down to just adds and converts in the loop?
#floats #n9
BUG=skia:3592
CQ_INCLUDE_TRYBOTS=client.skia.android:Test-Android-Nexus9-TegraK1-Arm64-Release-Trybot
Review URL: https://codereview.chromium.org/1008973004
The xps gyp target depends on skia_lib, which in turn contains
the sfnt target (and re-exports it settings). As a result, it
should not separately depend on the sfnt target. This currently
isn't causing issues because the sfnt target is mostly header only,
but any code in it may be duplicated.
Review URL: https://codereview.chromium.org/1038693003
Updates canDraw to accept all the same Skia/Gr objects as the drawText
functions, since that information may very well be relevant in
determining whether a context can draw.
Also moves the onDrawTextBlob implementation directly into
drawTextBlob.
BUG=skia:
Review URL: https://codereview.chromium.org/1010113004
There is no reason to require the 4 SkPMFloats (registers) to be adjacent.
The only potential win in loads and stores comes from the SkPMColors being adjacent.
Makes no difference to existing bench.
BUG=skia:
Review URL: https://codereview.chromium.org/1035583002
