This patch adds per-benchmark-iteration times to our JSON output. Given that we
already collect these statistics, giving them to the user would be nice.
No unit-test provided, since `rgrep -i json tests` yielded nothing. Happy to add
one if someone wants.
BUG=None.
TEST=nanobench now writes per-run timinings with the output JSON.
Change-Id: I910f1d97fd3e0ee69fc8e78e011e67b9c866f18d
Reviewed-on: https://skia-review.googlesource.com/5617
Reviewed-by: Mike Klein <mtklein@chromium.org>
Commit-Queue: Ravi Mistry <rmistry@google.com>
This will hopefully make a later change that refactors how aa is passed to GrRenderTargetContext a little easier to review.
Change-Id: Ie750d237714a0407ec1e5604c0daa080da32519b
Reviewed-on: https://skia-review.googlesource.com/5630
Reviewed-by: Matt Sarett <msarett@google.com>
Commit-Queue: Brian Salomon <bsalomon@google.com>
BUG=skia:
Change-Id: I8866df425ee9837e75f0b2f76777f7e5d68fb21d
Reviewed-on: https://skia-review.googlesource.com/5624
Commit-Queue: Brian Osman <brianosman@google.com>
Reviewed-by: Mike Klein <mtklein@chromium.org>
1) rename to bilerp_xy, for x,y in {n[egative], p[ositive};
2) pull out a save_xy stage to save off the original x,y;
3) also calculate the fractional x,y fx,fy once instead of 4 times.
1) is a pure refactor;
2) adds a stage but otherwise is nothing different;
3) changes images a little bit (fractional parts can vary a bit around powers of two).
This extends naturally to naive bicubic using 16 bicubic_xy stages.
CQ_INCLUDE_TRYBOTS=skia.primary:Test-Ubuntu-GCC-GCE-CPU-AVX2-x86_64-Release-SKNX_NO_SIMD
Change-Id: I666de5c21e978abb4feb6e3225e5b5920ba6c5b9
Reviewed-on: https://skia-review.googlesource.com/5550
Commit-Queue: Mike Klein <mtklein@chromium.org>
Reviewed-by: Herb Derby <herb@google.com>
Split up DEF_TEST(Codec, r) by type. Frequently when I am testing
locally I only want to test one type (e.g. I am testing changes to
SkGifCodec), and this allows me to skip past the ones I do not care
about.
It also speeds up my local runs, since most of the time is spent in
this single long test. Splitting them up means they can run in
parallel, halving run time. (Probably doesn't make a big difference
on bots, where more tests are running.)
Change-Id: Ic6babc20e1288dbb4fb255290f71523a033c7559
Reviewed-on: https://skia-review.googlesource.com/5622
Commit-Queue: Leon Scroggins <scroggo@google.com>
Reviewed-by: Matt Sarett <msarett@google.com>
This is already on the bitmap provider - no need to plumb it here.
BUG=skia:
Change-Id: Ie8c5aef8ea443c58d59598aa66e357ad4c212b25
Reviewed-on: https://skia-review.googlesource.com/5621
Commit-Queue: Brian Osman <brianosman@google.com>
Reviewed-by: Mike Klein <mtklein@chromium.org>
There's a bit of friction with this attribute, because per spec it is
an inherited presentation attribute, but in Skia it is part of the
actual SkPath state.
So we must add some plumbing to SkSVGShape & friends to allow overriding
the fill type at render-time.
R=robertphillips@google.com,stephana@google.com
Change-Id: I9c926d653c6211beb3914bffac50d4349dbdd2c0
Reviewed-on: https://skia-review.googlesource.com/5415
Reviewed-by: Mike Reed <reed@google.com>
Commit-Queue: Florin Malita <fmalita@chromium.org>
CQ_INCLUDE_TRYBOTS=skia.primary:Test-Ubuntu-Clang-GCE-CPU-AVX2-x86_64-Debug-ASAN,Test-Ubuntu-GCC-GCE-CPU-AVX2-x86_64-Release-SKNX_NO_SIMD
Change-Id: I6aef3ec2bcb85d6ad4830c759a1234ef2e31f32d
Reviewed-on: https://skia-review.googlesource.com/5552
Commit-Queue: Mike Klein <mtklein@chromium.org>
Reviewed-by: Mike Klein <mtklein@chromium.org>
Bicubic is going to blow right past 48. At this point the fixed preallocation strategy is starting to look naive... at 64 we'd allocate just over 1K for every pipeline (and every compiled pipeline).
CQ_INCLUDE_TRYBOTS=skia.primary:Test-Ubuntu-GCC-GCE-CPU-AVX2-x86_64-Release-SKNX_NO_SIMD
Change-Id: Ib2944ead1217123aba2b6347fd9d5315217540c9
Reviewed-on: https://skia-review.googlesource.com/5551
Reviewed-by: Mike Klein <mtklein@chromium.org>
Commit-Queue: Mike Klein <mtklein@chromium.org>
It appears to be confusing "git cl upload".
BUG=skia:6033
Change-Id: I9813798dbfbbdb9eba6f4159f07e040ac865006c
Reviewed-on: https://skia-review.googlesource.com/5601
Reviewed-by: Eric Boren <borenet@google.com>
Commit-Queue: Ravi Mistry <rmistry@google.com>
The GN in the name only matters for iOS bots now. All the others are GN-only.
This revealed a couple redundant build bots: we had two each of Win x86_64 Debug and Release builders, one with -GN and one not.
I have intentionally not resorted gen_tasks.go to keep the diff looking somewhat sane. I'd be happy to sort here or to follow up with a re-sort CL.
Change-Id: I2f8a136a1fbde416739966e27685b52b94b50cf6
Reviewed-on: https://skia-review.googlesource.com/5545
Reviewed-by: Eric Boren <borenet@google.com>
FT_LOAD_BITMAP_METRICS_ONLY flag was introduced for retrieving font
metrics without decoding or allocating bitmaps.
BUG=skia:
Change-Id: I901531501111f24d8b670305379e04c0bc688e6f
Reviewed-on: https://skia-review.googlesource.com/5580
Reviewed-by: Ben Wagner <bungeman@google.com>
Commit-Queue: Ben Wagner <bungeman@google.com>
BUG=skia:
Change-Id: Ib1920fffd5735ad54a5b785bbc2676ea240bdbfa
Reviewed-on: https://skia-review.googlesource.com/5611
Reviewed-by: Mike Klein <mtklein@chromium.org>
Commit-Queue: Brian Osman <brianosman@google.com>
SkGIFFrameContext::decode() and SkGIFLZWContext::prepareToDecode() do
not need (or use) the global color map, so stop passing it as a
parameter. The parameter was used prior to
https://skia-review.googlesource.com/c/4379/ (different issue!), but we
overlooked removing it then.
Change-Id: I0f477e9db11f7650938d6b868baef69e3b37d86b
Reviewed-on: https://skia-review.googlesource.com/5609
Commit-Queue: Leon Scroggins <scroggo@google.com>
Reviewed-by: Matt Sarett <msarett@google.com>
Currently, just inject the Ganesh context type when running unit tests.
Obviously, we can use this to supply other contextual information around
tests that do many variations of configs, formats, etc...
BUG=skia:
Change-Id: Iab96632a92ec632e4d132bbcc17a91a8dd251e78
Reviewed-on: https://skia-review.googlesource.com/5565
Reviewed-by: Brian Salomon <bsalomon@google.com>
Commit-Queue: Brian Osman <brianosman@google.com>
SkColorSpace::MakeICC now parses Gray ICC profiles and
SkColorSpaceXform_A2B can now render color spaces from Gray ICC
profiles. This is not enabled for SkPngCodec as of right now as we don't
have any Gray PNG test images currently.
BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=5214
CQ_INCLUDE_TRYBOTS=skia.primary:Test-Ubuntu-GCC-GCE-CPU-AVX2-x86_64-Release-SKNX_NO_SIMD
Change-Id: Iea7136d1c163cd17cf0818af57f93efdbcb577e7
Reviewed-on: https://skia-review.googlesource.com/5214
Reviewed-by: Matt Sarett <msarett@google.com>
Commit-Queue: Robert Aftias <raftias@google.com>
BUG=skia:
Change-Id: Id66a68e47f671cafd7c6128787fbd149faf16c7a
Reviewed-on: https://skia-review.googlesource.com/5576
Reviewed-by: Brian Salomon <bsalomon@google.com>
Commit-Queue: Brian Osman <brianosman@google.com>
The code in views and samplecode wasn't being compiled in. Expand the
scope of the SK_ANGLE define to fix that. Add some missing deps too.
BUG=skia:
Change-Id: Id7b584457a21f171efc2339fb969c550379295e5
Reviewed-on: https://skia-review.googlesource.com/5575
Reviewed-by: Mike Klein <mtklein@chromium.org>
Commit-Queue: Mike Klein <mtklein@chromium.org>
SkNoDrawCanvas is not backed by pixels, but for draw ops not intercepted
by clients we abort rasterization failry late (SkNullBlitter).
BUG=chromium:668925
R=reed@google.com
Change-Id: I4cd80dbbc262936d33410275051ea0b9c04fbc6c
Reviewed-on: https://skia-review.googlesource.com/5543
Commit-Queue: Florin Malita <fmalita@google.com>
Reviewed-by: Mike Reed <reed@google.com>
This is a follow-up to reviews.skia.org/5540, which did float -> byte.
We use the same trick here exploiting 32768.0f / 0x47000000.
The benefit here is smaller than the other CL, but still measurable.
The exchange here is:
before: int->float, multiply
after: OR, FMA
The cost of an FMA is the same as a multiply, so we're basically just replacing int->float conversion with a bitwise OR.
CQ_INCLUDE_TRYBOTS=skia.primary:Test-Ubuntu-GCC-GCE-CPU-AVX2-x86_64-Release-SKNX_NO_SIMD
Change-Id: Ieac2247664afa3ff415aec2b48c21505905bee23
Reviewed-on: https://skia-review.googlesource.com/5542
Reviewed-by: Matt Sarett <msarett@google.com>
Commit-Queue: Mike Klein <mtklein@chromium.org>
This reverts commit 155b29224d.
Reason for revert:
https://skia-review.googlesource.com/c/5541/ (Second part of piecemeal revert) seems to have zeroed in on the change that caused the perf regression. Let's see if the changes reverted in this CL are acceptable.
Change-Id: I477fe23d84a9c14f87a431ec2b495f617ff410be
Reviewed-on: https://skia-review.googlesource.com/5573
Commit-Queue: Robert Phillips <robertphillips@google.com>
Reviewed-by: Robert Phillips <robertphillips@google.com>
BUG=skia:
Change-Id: I9ef5de7c94377ef54a10e14efa59f7e142543da6
Reviewed-on: https://skia-review.googlesource.com/5571
Reviewed-by: Ravi Mistry <rmistry@google.com>
Commit-Queue: Eric Boren <borenet@google.com>
Add a test
BUG=skia:3534
BUG=b/33300701
Change-Id: Ifb3a824a36998c5e626c4ad58466845f49d18ebf
Reviewed-on: https://skia-review.googlesource.com/5568
Commit-Queue: Leon Scroggins <scroggo@google.com>
Reviewed-by: Matt Sarett <msarett@google.com>
This is a partial revert of https://codereview.chromium.org/2514543002 (Defer more renderTargetContexts in the GPU image filter paths - take 2)
I have been unable to reproduce the performance regression in crbug.com/668179 locally so intend to revert the above CL piecemeal.
BUG=668179
Change-Id: Iee9d0164f85ae33ff8dfa9b3eb01c07fd825f017
Reviewed-on: https://skia-review.googlesource.com/5541
Reviewed-by: Brian Salomon <bsalomon@google.com>
Commit-Queue: Robert Phillips <robertphillips@google.com>
In IEEE, for each byte BB, the float 0x470000BB equals 32768.0f + BB*(1/256.0f).
So to turn a [0,1] float into a byte, we can
- multiply by (255/256.0f) to get into [0,255/256.0f] range,
- add 32768.0f to get into [32768.0f, 32768.0f + 255/256.0f] range,
- look at the low byte.
Those first two of course are an FMA.
Using this trick here makes store_8888 measurably faster. Instead of a FMA then float->int trunc, we do an FMA then a bitwise AND. Overall the math goes from 4 FMA + 4 trunc + 3 shift to 4 FMA + 3 AND + 3 shift (we can skip the shift for red and the AND for alpha). As you might guess, AND is cheaper than trunc, so this is a net win.
I should be able to follow up with the same trick in reverse in from_8888().
CQ_INCLUDE_TRYBOTS=skia.primary:Test-Ubuntu-GCC-GCE-CPU-AVX2-x86_64-Release-SKNX_NO_SIMD
Change-Id: I42c8f4a6ea0b6c22160517cf5f9c048f01c9a330
Reviewed-on: https://skia-review.googlesource.com/5540
Reviewed-by: Matt Sarett <msarett@google.com>
Commit-Queue: Mike Klein <mtklein@chromium.org>
The new values are calculated starting with the "true" values,
adapted to D50 using the Bradford transform.
I don't remember where the old values came from, they've been
around forever. It's possible that I just typed in the values
that came out of ICC profiles (these would have been imperfect,
given that there's a fixed-point -> float conversion happening
there).
BUG=skia:
Change-Id: I8e870266bc8ab5372c3ccf6a5ee6691b2915af43
Reviewed-on: https://skia-review.googlesource.com/5450
Commit-Queue: Matt Sarett <msarett@google.com>
Reviewed-by: Brian Osman <brianosman@google.com>
Two of my previous CLs stepped on each other.
BUG=skia:
Change-Id: I53d49f283e466ad7fd6c38a6d4243c084dd324c7
Reviewed-on: https://skia-review.googlesource.com/5519
Reviewed-by: Eric Boren <borenet@google.com>
We were trying to mix the canvas' color space (linear gamma) with N32
color type, which isn't allowed. Propagate the original color type, too.
BUG=skia:
Change-Id: I606b25e690486abe042d6a4b6dda606ac85b546b
Reviewed-on: https://skia-review.googlesource.com/5509
Commit-Queue: Brian Osman <brianosman@google.com>
Reviewed-by: Brian Salomon <bsalomon@google.com>
BUG=skia:
Change-Id: I40d46e3a8a5c6c6fa77075c94d6bd290f764a41c
Reviewed-on: https://skia-review.googlesource.com/5512
Reviewed-by: Ravi Mistry <rmistry@google.com>
Commit-Queue: Eric Boren <borenet@google.com>
Rename Win->Win10 where appropriate
BUG=skia:
Change-Id: I8692340cebf646e2cb0cb825e808b5ea1b695a8f
Reviewed-on: https://skia-review.googlesource.com/5356
Commit-Queue: Eric Boren <borenet@google.com>
Reviewed-by: Mike Klein <mtklein@chromium.org>
Reviewed-by: Ben Wagner <benjaminwagner@google.com>
BUG=skia:
Change-Id: I9a6a8970bd70dc52334fc44ed29ea942c29132fc
Reviewed-on: https://skia-review.googlesource.com/5507
Reviewed-by: Ravi Mistry <rmistry@google.com>
Commit-Queue: Eric Boren <borenet@google.com>
BUG=skia:5933
Change-Id: I440414d3d3db72a55be493aeb910bda29cc87841
Reviewed-on: https://skia-review.googlesource.com/5473
Reviewed-by: Brian Salomon <bsalomon@google.com>
Reviewed-by: Eric Boren <borenet@google.com>
Commit-Queue: Ben Wagner <benjaminwagner@google.com>
BUG:670620
Change-Id: Ic481d09a7112ef05f53fa1f94a7c155870c43408
Reviewed-on: https://skia-review.googlesource.com/5501
Commit-Queue: Matt Sarett <msarett@google.com>
Reviewed-by: Brian Osman <brianosman@google.com>
Reviewed-by: Jeremy Roman <jbroman@chromium.org>
This fixes a compile error in Chromium.
BUG=skia:6026
Change-Id: Idd5ad22cb52a084836de6e1427f1f047d1feab08
Reviewed-on: https://skia-review.googlesource.com/5500
Reviewed-by: Derek Sollenberger <djsollen@google.com>
Commit-Queue: Leon Scroggins <scroggo@google.com>
1) Our older iOS devices failed our sRGB tests, due to precision issues
with alpha. At this point, we only test on iPadMini 4, and that appears
not to have any problems.
2) iOS devices still don't have the sRGB texture decode extension. But,
some clients have no interest in mixing legacy/color-correct rendering,
and would like to use sRGB on these devices. This GrContextOptions flag
enables sRGB support in those cases.
Adjust the test code to produce sRGB capable contexts on these devices,
but only for configs that have a color space. (See comment).
BUG=skia:4148
Committed: https://skia.googlesource.com/skia/+/9db12d2341f3f8722c8b90b11dd4cce138a8a64e
Committed: https://skia.googlesource.com/skia/+/1aeb78c5d978b35b256525b711edd942bce01444
Review-Url: https://codereview.chromium.org/2539993002
Fixes many blatant errors with glnarrow config
BUG=skia:
Change-Id: I729cda350ebce126bdac4eb41c91f30294e1c61e
Reviewed-on: https://skia-review.googlesource.com/5469
Reviewed-by: Mike Klein <mtklein@chromium.org>
Commit-Queue: Brian Osman <brianosman@google.com>
- nothing uses ashmem;
- cpufeatures and native_app_glue are now pulled from the NDK;
- no bots use the scripts in platform_tools/android/bin;
- update scripts to work with GN instead of GYP.
Change-Id: I14f47eeadb3047505e232dd10385f58ef12c73f4
Reviewed-on: https://skia-review.googlesource.com/5422
Reviewed-by: Mike Klein <mtklein@chromium.org>
Commit-Queue: Derek Sollenberger <djsollen@google.com>
This gives us a place to bottleneck this sort of conversion. Every time I try to use the rounding float -> int instructions, they're just a little slower than working the 1/2 into the scale with FMA. Weird.
Change-Id: I7718112b234b4b38ba6af8fef59a47642021839a
Reviewed-on: https://skia-review.googlesource.com/5483
Commit-Queue: Mike Klein <mtklein@chromium.org>
Reviewed-by: Matt Sarett <msarett@google.com>
I think we just happened not to here. This improves Adobe -> sRGB pipeline conversion by about 3-4%.
While at it, unify all the fma() lambdas into SkNf_fma(). I'd have called it fma(), but IIRC there was some sort of name conflict there with type-generic fma() functions from the C math.h or something silly like that.
CQ_INCLUDE_TRYBOTS=skia.primary:Test-Ubuntu-GCC-GCE-CPU-AVX2-x86_64-Release-SKNX_NO_SIMD
Change-Id: Id176671fec27c984efa4703c5be2fb63b7f0b11f
Reviewed-on: https://skia-review.googlesource.com/5474
Reviewed-by: Matt Sarett <msarett@google.com>
Commit-Queue: Mike Klein <mtklein@chromium.org>
Change-Id: I3ebe5a471477ce1b71c150b0bde4982d113fd8a7
Reviewed-on: https://skia-review.googlesource.com/5468
Reviewed-by: Brian Osman <brianosman@google.com>
Commit-Queue: Brian Salomon <bsalomon@google.com>
