The existing bench only tests the fast path, but we're looking to speed
up the general case. It'd be nice to be able to measure that speedup.
BUG=skia:
Review URL: https://codereview.chromium.org/1146953003
Constructing the gm tests and benches causes many calls to font loads.
This is visible as profiling samples in fontconfig and freetype on Linux
for all profiling runs of nanobench. This complicates analysis of
test-cases that are suspected of being slow due to font-related issues.
Move the font loading to GM::onOnceBeforeDraw and Benchmark::onPreDraw.
This way the code is not executed if the testcase does not match the
nanobench --match filter. This way the samples in font-related code are
more easy to identify as legitimate occurances caused by the testcase.
This should not cause differences in timings, because:
* Benchmark::preDraw / onPreDraw is defined to be run outside the timer
* GM::runAsBench is not enabled for any of the modified testcases. Also
nanobench untimed warmup round should run the onOnceBeforeDraw.
(and there are other GM::runAsBench gms already doing loading in
onOnceBeforeDraw).
Changes the behavior:
In TextBench:
Before, the test would report two different gms with the same name if
the color emoji font was not loaded successfully.
After, the test always reports all tests as individual names.
Generally:
The errors from loading fonts now print inbetween each testcase, as
opposed to printing during construction phase. Sample output:
( 143/145 MB 1872) 14.7ms 8888 gm quadclosepathResource /fonts/Funkster.ttf not a valid font.
( 160/160 MB 1831) 575µs 8888 gm surfacenewResource /fonts/Funkster.ttf not a valid font.
( 163/165 MB 1816) 12.5ms 8888 gm linepathResource /fonts/Funkster.ttf not a valid font.
( 263/411 MB 1493) 118ms 8888 gm typefacestyles_kerningResource /fonts/Funkster.ttf not a valid font.
( 374/411 MB 1231) 7.16ms 565 gm getpostextpathResource /fonts/Funkster.ttf not a valid font.
( 323/411 MB 1179) 4.92ms 565 gm stringartResource /fonts/Funkster.ttf not a valid font.
( 347/493 MB 917) 191ms 565 gm patch_gridResource /fonts/Funkster.ttf not a valid font.
( 375/493 MB 857) 23.9ms gpu gm clipdrawdrawCannot render path (0)
( 393/493 MB 706) 2.91ms unit test ParsePath------ png error IEND: CRC error
( 394/493 MB 584) 166ms gpu gm hairmodesResource /fonts/Funkster.ttf not a valid font.
Resource /fonts/Funkster.ttf not a valid font.
Resource /fonts/Funkster.ttf not a valid font.
...
Review URL: https://codereview.chromium.org/1144023002
Make GrResourceCache performance less sensitive to key length change.
The memcmp in GrResourceKey is called when SkTDynamicHash jumps the
slots to find the hash by a index. Avoid most of the memcmps by
comparing the hash first.
This is important because small changes in key data length can cause
big performance regressions. The theory is that key length change causes
different hash values. These hash values might trigger memcmps that
originally weren't there, causing the regression.
Adds few specialized benches to grresourcecache_add to test different
key lengths. The tests are run only on release, because on debug the
SkTDynamicHash validation takes too long, and adding many such delays
to development test runs would be unproductive. On release the tests
are quite fast.
Effect of this patch to the added tests on amd64:
grresourcecache_find_10 738us -> 768us 1.04x
grresourcecache_find_2 472us -> 476us 1.01x
grresourcecache_find_25 841us -> 845us 1x
grresourcecache_find_4 565us -> 531us 0.94x
grresourcecache_find_54 1.18ms -> 1.1ms 0.93x
grresourcecache_find_5 834us -> 749us 0.9x
grresourcecache_find_3 620us -> 542us 0.87x
grresourcecache_add_25 2.74ms -> 2.24ms 0.82x
grresourcecache_add_56 3.23ms -> 2.56ms 0.79x
grresourcecache_add_54 3.34ms -> 2.62ms 0.78x
grresourcecache_add_5 2.68ms -> 2.1ms 0.78x
grresourcecache_add_10 2.7ms -> 2.11ms 0.78x
grresourcecache_add_2 1.85ms -> 1.41ms 0.76x
grresourcecache_add 1.84ms -> 1.4ms 0.76x
grresourcecache_add_4 1.99ms -> 1.49ms 0.75x
grresourcecache_add_3 2.11ms -> 1.55ms 0.73x
grresourcecache_add_55 39ms -> 13.9ms 0.36x
grresourcecache_find_55 23.2ms -> 6.21ms 0.27x
On arm64 the results are similar.
On arm_v7_neon, the results lack the discontinuity at 55:
grresourcecache_add 4.06ms -> 4.26ms 1.05x
grresourcecache_add_2 4.05ms -> 4.23ms 1.05x
grresourcecache_find 1.28ms -> 1.3ms 1.02x
grresourcecache_find_56 3.35ms -> 3.32ms 0.99x
grresourcecache_find_2 1.31ms -> 1.29ms 0.99x
grresourcecache_find_54 3.28ms -> 3.24ms 0.99x
grresourcecache_add_5 6.38ms -> 6.26ms 0.98x
grresourcecache_add_55 8.44ms -> 8.24ms 0.98x
grresourcecache_add_25 7.03ms -> 6.86ms 0.98x
grresourcecache_find_25 2.7ms -> 2.59ms 0.96x
grresourcecache_find_4 1.45ms -> 1.38ms 0.95x
grresourcecache_find_10 2.52ms -> 2.39ms 0.95x
grresourcecache_find_55 3.54ms -> 3.33ms 0.94x
grresourcecache_find_5 2.5ms -> 2.32ms 0.93x
grresourcecache_find_3 1.57ms -> 1.43ms 0.91x
The extremely slow case, 55, is postulated to be due to the index jump
collisions running the memcmp. This is not visible on arm_v7_neon probably due
to hash function producing different results for 32 bit architectures.
This change is needed for extending path cache key in Gr
NV_path_rendering codepath. Extending is needed in order to add dashed
paths to the path cache.
Review URL: https://codereview.chromium.org/1132723003
I'm thinking of using this in perf with something like:
ratio(fill(filter("test=foo")), fill(filter("test=control")))
Does that make sense to you?
Not sure that this is really a good control bench on all bots,
but I propose we just run it a bit and find out if it needs work.
BUG=skia:
Review URL: https://codereview.chromium.org/1129823003
The benches for N <= 10 get around 2x faster on my N7 and N9. I believe this
is because of the reduced function-call-then-function-pointer-call overhead on
the N7, and additionally because it seems autovectorization beats our NEON code
for small N on the N9.
My desktop is unchanged, though that's probably because N=10 lies well within a
region where memset's performance is essentially constant: N=100 takes only
about 2x as long as N=1 and N=10, which perform nearly identically.
BUG=skia:
Review URL: https://codereview.chromium.org/1073863002
The colorfilter is applied to a single (paint's) color, so the bench does not
measure the filter at all, but simply the blit of a color.
BUG=skia:
TBR=
Review URL: https://codereview.chromium.org/1055383002
Now that all SkCodecs can rewind (assuming the stream is rewindable),
we do not need to special case it.
Pointed out by Derek in the code review that added this.
TBR=djsollen
Review URL: https://codereview.chromium.org/1058633002
CodecBench:
Add new class for timing using SkCodec.
DecodingBench:
Include creating a decoder inside the loop. This is to have a better
comparison against SkCodec. SkCodec's factory function does not
necessarily read the same amount as SkImageDecoder's, so in order to
have a meaningful comparison, read the entire stream from the
beginning. Also for comparison, create a new SkStream from the
SkData each time.
Add a debugging check to make sure we have an SkImageDecoder.
Add include guards.
nanobench.cpp:
Decode using SkCodec.
When decoding using SkImageDecoder, exclude benches where we decoded
to a different color type than requested. SkImageDecoder may decide to
decode to a different type, in which case the name is misleading.
TODOs:
Now that we ignore color types that do not match the desired
color type, we should add Index8. This also means calling the more
complex version of getPixels so CodecBench can support kIndex8.
BUG=skia:3257
Review URL: https://codereview.chromium.org/1044363002
The primary feature this delivers is SkNf and SkNd for arbitrary power-of-two N. Non-specialized types or types larger than 128 bits should now Just Work (and we can drop in a specialization to make them faster). Sk4s is now just a typedef for SkNf<4, SkScalar>; Sk4d is SkNf<4, double>, Sk2f SkNf<2, float>, etc.
This also makes implementing new specializations easier and more encapsulated. We're now using template specialization, which means the specialized versions don't have to leak out so much from SkNx_sse.h and SkNx_neon.h.
This design leaves us room to grow up, e.g to SkNf<8, SkScalar> == Sk8s, and to grown down too, to things like SkNi<8, uint16_t> == Sk8h.
To simplify things, I've stripped away most APIs (swizzles, casts, reinterpret_casts) that no one's using yet. I will happily add them back if they seem useful.
You shouldn't feel bad about using any of the typedef Sk4s, Sk4f, Sk4d, Sk2s, Sk2f, Sk2d, Sk4i, etc. Here's how you should feel:
- Sk4f, Sk4s, Sk2d: feel awesome
- Sk2f, Sk2s, Sk4d: feel pretty good
No public API changes.
TBR=reed@google.com
BUG=skia:3592
Review URL: https://codereview.chromium.org/1048593002
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
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
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
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
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
RotatedRectBench was asking for its base layer size, which may
not be what it expects with odd canvas modes (particularly proxies).
Most benchmarks are not so sophisticated; they hard-wire their
size and just use that (expected) value.
R=mtklein@google.com,djsollen@google.com
BUG=skia:3566
Review URL: https://codereview.chromium.org/1015013004
Initial experiments did show that the 256 tile size fixed the hd2000 win7
nanobot failures. However it did not have any effect on other bots, so this
change is to move back to the larger tile size on all bots expect for the
hd2000.
BUG=skia:
Review URL: https://codereview.chromium.org/1022083002
Going back to old nanobench tile size to see if the increase to tile is what has been
causing recent nanobench crashes. The crashes seem very nondeterministic and hard to
debug manually.
256x256 is too small of a tile to give accurate gpu results but if this fixes we can try some compromise in the middle
BUG=skia:
Review URL: https://codereview.chromium.org/1022823003
(This is essentially a revert of https://codereview.chromium.org/503833002/.)
This was necessary back when SkPaint was flattened even for in-process use. Now that we only flatten SkPaint for cross-process use, there's no need to serialize UniqueIDs.
Note: SkDropShadowImageFilter is being constructed with a croprect and UniqueID (of 0) in Blink. I've made the uniqueID param default to 0 temporarily, until this rolls in and Blink can be changed. (Blink can't be changed first, since unlike the other filters, there's no constructor that takes a cropRect but not a uniqueID.)
BUG=skia:
Review URL: https://codereview.chromium.org/1019493002
Seems strictly more useful.
This implements Mac and Windows, which seemed easy. Don't know how to do this on Linux yet.
BUG=skia:
CQ_EXTRA_TRYBOTS=client.skia:Test-Mac10.9-MacMini6.2-HD4000-x86_64-Debug-Trybot
NOTREECHECKS=true
TBR=halcanary@google.com
Review URL: https://codereview.chromium.org/990723002
Please see if this looks usable. It may even give a perf boost if you use it, even without custom implementations for each instruction set.
I've been trying this morning to beat this naive loop implementation, but so far no luck with either _SSE2.h or _SSSE3.h. It's possible this is an artifact of the microbenchmark, because we're not doing anything between the conversions. I'd like to see how this fits into real code, what assembly's generated, what the hot spots are, etc.
I've updated the tests to test these new APIs, and splintered off a pair of new benchmarks that use the new APIs. This required some minor rejiggering in the benches.
BUG=skia:
Review URL: https://codereview.chromium.org/978213003
Instead of set(SkPMColor), add a constructor SkPMFloat(SkPMColor).
Replace setA(), setR(), etc. with a 4 float constructor.
And, promise to stick to SkPMColor order.
BUG=skia:
Review URL: https://codereview.chromium.org/977773002
This bench was ~75% overhead, ~25% good bench. It is now just about the
opposite: about 30% of the runtime is loop and random number overhead, and
about 70% of the time is spent doing SkPMColor <-> SkPMFloat work.
BUG=skia:
NOPRESUBMIT=true
Review URL: https://codereview.chromium.org/968133005
BUG=skia:1366
For the added bench, the collapsing makes the bench take:
- 70% of the time for CPU rendering of 3 consecutive matrix filters
- almost no change in the GPU rendering of the matrix filters
- 50% of the time for CPU and GPU rendering of 3 consecutive table filters
Review URL: https://codereview.chromium.org/776673002
This basically takes out the Windows-only hacks and promotes them to
cross-platform behavior driven by --gpu_threading.
- When --gpu_threading is false (the default), this puts GPU tasks and tests
together in the same GPU enclave. They all run serially.
- When --gpu_threading is true, both the tests and the tasks run totally
independently, just like the thread-safe CPU-bound work.
BUG=skia:3255
Review URL: https://codereview.chromium.org/847273005
SkStream is a stateful object, so it does not make sense for it to have
multiple owners. Make SkStream inherit directly from SkNoncopyable.
Update methods which previously called SkStream::ref() (e.g.
SkImageDecoder::buildTileIndex() and SkFrontBufferedStream::Create(),
which required the existing owners to call SkStream::unref()) to take
ownership of their SkStream parameters and delete when done (including
on failure).
Switch all SkAutoTUnref<SkStream>s to SkAutoTDelete<SkStream>s. In some
cases this means heap allocating streams that were previously stack
allocated.
Respect ownership rules of SkTypeface::CreateFromStream() and
SkImageDecoder::buildTileIndex().
Update the comments for exceptional methods which do not affect the
ownership of their SkStream parameters (e.g.
SkPicture::CreateFromStream() and SkTypeface::Deserialize()) to be
explicit about ownership.
Remove test_stream_life, which tested that buildTileIndex() behaved
correctly when SkStream was a ref counted object. The test does not
make sense now that it is not.
In SkPDFStream, remove the SkMemoryStream member. Instead of using it,
create a new SkMemoryStream to pass to fDataStream (which is now an
SkAutoTDelete).
Make other pdf rasterizers behave like SkPDFDocumentToBitmap.
SkPDFDocumentToBitmap delete the SkStream, so do the same in the
following pdf rasterizers:
SkPopplerRasterizePDF
SkNativeRasterizePDF
SkNoRasterizePDF
Requires a change to Android, which currently treats SkStreams as ref
counted objects.
Review URL: https://codereview.chromium.org/849103004
Restructure SkGpuDevice creation:
*SkSurfaceProps are optional.
*Use SkSurfaceProps to communicate DF text rather than a flag.
*Tell SkGpuDevice::Create whether RT comes from cache or not.
Review URL: https://codereview.chromium.org/848903004
BUG=skia:3255
I think this supports everything DM used to, but has completely refactored how
it works to fit the design in the bug.
Configs like "tiles-gpu" are automatically wired up.
I wouldn't suggest looking at this as a diff. There's just a bunch of deleted
files, a few new files, and one new file that shares a name with a deleted file
(DM.cpp).
NOTREECHECKS=true
Committed: https://skia.googlesource.com/skia/+/709d2c3e5062c5b57f91273bfc11a751f5b2bb88
Review URL: https://codereview.chromium.org/788243008
Reason for revert:
plenty of data
Original issue's description:
> Sketch DM refactor.
>
> BUG=skia:3255
>
>
> I think this supports everything DM used to, but has completely refactored how
> it works to fit the design in the bug.
>
> Configs like "tiles-gpu" are automatically wired up.
>
> I wouldn't suggest looking at this as a diff. There's just a bunch of deleted
> files, a few new files, and one new file that shares a name with a deleted file
> (DM.cpp).
>
> NOTREECHECKS=true
>
> Committed: https://skia.googlesource.com/skia/+/709d2c3e5062c5b57f91273bfc11a751f5b2bb88TBR=bsalomon@google.com,mtklein@chromium.org
NOTREECHECKS=true
NOTRY=true
BUG=skia:3255
Review URL: https://codereview.chromium.org/853883004
BUG=skia:3255
I think this supports everything DM used to, but has completely refactored how
it works to fit the design in the bug.
Configs like "tiles-gpu" are automatically wired up.
I wouldn't suggest looking at this as a diff. There's just a bunch of deleted
files, a few new files, and one new file that shares a name with a deleted file
(DM.cpp).
NOTREECHECKS=true
Review URL: https://codereview.chromium.org/788243008
This avoids the problem of a newly created uncached texture causing a purge of cached resources.
BUG=chromium:445885
Review URL: https://codereview.chromium.org/846303002
This fixes every case where virtual and SK_OVERRIDE were on the same line,
which should be the bulk of cases. We'll have to manually clean up the rest
over time unless I level up in regexes.
for f in (find . -type f); perl -p -i -e 's/virtual (.*)SK_OVERRIDE/\1SK_OVERRIDE/g' $f; end
BUG=skia:
Review URL: https://codereview.chromium.org/806653007
It is desirable that, when layer hoisting is disabled, the MPD and non-MPD timings be
roughly the same. Unfortunately, using a separate canvas for each tile (a requirement
for MPD) introduces its own discrepancy into the timing. Using a separate canvas for
each tile doesn't seem to make a difference for 8888 (see the non-MPD 8888 column below)
but slows down GPU rendering (see the non-MPD GPU column below). Since this is how
Chromium renders I propose switching to this regimen (even though it is "slowing down"
GPU rendering).
nanobench mean times (ms) with layer hoisting disabled (for desk_amazon.skp)
8888
MPD non-MPD
1 canvas (old-style) 0.628 1.71
separate (new-style) 0.795 1.63
GPU
MPD non-MPD
1 canvas (old-style) 2.34 1.69
separate (new-style) 2.32 2.66
Review URL: https://codereview.chromium.org/779643002
Two issues with the SKPBench tile computation were causing the MPD path to do more work:
The clip from the parent canvas wasn't being used to trim content off the edges of the MPD tiles
The non-MPD path was not taking the scale into account in its tile placement (resulting in it having fewer, larger active tiles when scaling).
Review URL: https://codereview.chromium.org/776273002
Seems okay after this small patch to skip lockPixels() / unlockPixels().
BUG=skia:3149
CQ_EXTRA_TRYBOTS=client.skia:Test-Ubuntu13.10-GCE-NoGPU-x86_64-Release-TSAN-Trybot
Review URL: https://codereview.chromium.org/773203003
Removes work done by the constructors of picture_nesting benches,
and moves the work to the Benchmark::onPreDraw override.
This avoids PictureNesting::sierpinsky showing up in profile traces
when profiling other benches.
Review URL: https://codereview.chromium.org/725523002
Reason for revert:
Causing breakages on Mac build.
Original issue's description:
> Make nanobench and dm be usable from Chromium build
>
> Move the app logic for each app as follows:
>
> <app>.cpp -- the file which contains main(). Embedders that compile
> their own apps, such as ios shell, upcoming Chromium dm etc, do not use this.
>
> <app>_main.cpp -- the main logic of the Skia test application. This will be
> used by Skia -compiled apps as well as embedder -compiled apps.
>
> <app>_main.h -- the API for the main logic. This will be
> used by Skia -compiled apps as well as embedder -compiled apps.
>
> This way (the upcoming) Chromium dm can setup its Chromium-specific setup
> in custom main(), and then call dm_main(), without the need of any
> SK_BUILD_FOR_XXXX defines controlling whether the tool defines main or not.
>
> BUG=skia:2992
>
> Committed: https://skia.googlesource.com/skia/+/c092d3bdab5f723576cc0346cea3ee282a9cb444TBR=mtklein@chromium.org,mtklein@google.com,borenet@google.com,kkinnunen@nvidia.com
NOTREECHECKS=true
NOTRY=true
BUG=skia:2992
Review URL: https://codereview.chromium.org/724073002
Move the app logic for each app as follows:
<app>.cpp -- the file which contains main(). Embedders that compile
their own apps, such as ios shell, upcoming Chromium dm etc, do not use this.
<app>_main.cpp -- the main logic of the Skia test application. This will be
used by Skia -compiled apps as well as embedder -compiled apps.
<app>_main.h -- the API for the main logic. This will be
used by Skia -compiled apps as well as embedder -compiled apps.
This way (the upcoming) Chromium dm can setup its Chromium-specific setup
in custom main(), and then call dm_main(), without the need of any
SK_BUILD_FOR_XXXX defines controlling whether the tool defines main or not.
BUG=skia:2992
Review URL: https://codereview.chromium.org/657373002
The tests path_hairline_{small,big}_AA_conic were calling the test
function with NVPR. This caused a warning in nanobench.
The here removed hunk comes from commit referring to skia:2042 ("Enable
NVPR by default"). This is a workaround for a bug. The bug is fixed by
the commit referring to skia:2078 ("Logan bot fails NVPR assertion in
bench").
The proper fix is indeed make sure that path renderer chain ends up
trying software path renderer, if the path contains conics and is a
hairline.
The removed hunk refers also to skia:2033 ("Figure out what is happening
with conic path segments in NVPR"). The above solution is correct also in case
NVPR would support conics, as NVPR would not still support hairlines.
BUG=skia:2078
Review URL: https://codereview.chromium.org/685213005
Reason for revert:
Not compiling in ANGLE build
Original issue's description:
> Get gpudft support working in dm, gm, nanobench and bench_pictures
>
> Adds a new config to test distance field text.
> Clean up some flags and #defines to read "distance field text",
> not "distance field fonts" to be consistent with Chromium
>
> NOTREECHECKS=true
>
> Committed: https://skia.googlesource.com/skia/+/06ba179838ba4fe187cf290750aeeb4a02a2960bTBR=bsalomon@google.com,mtklein@google.com,reed@google.com
NOTREECHECKS=true
NOTRY=true
Review URL: https://codereview.chromium.org/707723005
Adds a new config to test distance field text.
Clean up some flags and #defines to read "distance field text",
not "distance field fonts" to be consistent with Chromium
NOTREECHECKS=true
Review URL: https://codereview.chromium.org/699453005
- The expected case is now a single bulk-load insert() call instead of N;
- reserve() and flushDeferredInserts() can fold into insert() now;
- SkBBH subclasses may take ownership of the bounds
This appears to be a performance no-op on both my Mac and N5. I guess
even the simplest indirect branch predictor ("same as last time") can predict
the repeated virtual calls to SkBBH::insert() perfectly.
BUG=skia:
Review URL: https://codereview.chromium.org/670213002
