|
35cd7332
|
2018-12-02T12:03:33
|
|
Refactor test shader style.
This change enforces a lot more consistency. We pass const char * to
the Compile functions instead of std::string. Also fixes the
indentation of C++11 block comments to be more consistent.
Bug: angleproject:2995
Change-Id: Id6e5ea94055d8cbd420df4ea2e81b2d96cb5ce78
Reviewed-on: https://chromium-review.googlesource.com/c/1357103
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Commit-Queue: Jamie Madill <jmadill@chromium.org>
|
|
|
b980c563
|
2018-11-27T11:34:27
|
|
Reformat all cpp and h files.
This applies git cl format --full to all ANGLE sources.
Bug: angleproject:2986
Change-Id: Ib504e618c1589332a37e97696cdc3515d739308f
Reviewed-on: https://chromium-review.googlesource.com/c/1351367
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Shahbaz Youssefi <syoussefi@chromium.org>
|
|
|
0e604f9f
|
2018-11-19T21:33:25
|
|
ANGLE Perf Test fixes.
Includes the following fixes:
* Removes obsolete global variable.
* Disables trace event collection when trace events disabled.
* Fixes calibrated steps getting stuck after first test.
* Scale calibrated steps with actual time after glFinish.
Should fix a few crashes and timeouts with the perf tests.
Bug: angleproject:2923
Change-Id: I9ba1c042dee31cf2400ccbeedca0d497ed52fb12
Reviewed-on: https://chromium-review.googlesource.com/c/1343677
Reviewed-by: Yuly Novikov <ynovikov@chromium.org>
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Commit-Queue: Jamie Madill <jmadill@chromium.org>
|
|
|
f3acb8c1
|
2018-11-14T16:24:25
|
|
Retool perf test running.
This change does a few things involving the way we run the perf tests:
- Perf test runner can capture different metrics
Useful for capturing the new "nanoSecPerIteration" metric.
- Removes the "score" metric
We'll move to the new time-based metrics. These new metrics are scaled
correctly with iteration counts.
- Runs three trials per perf test
This gives more measurements per test. Each trial is approximately one
second. First the perf tests set a fixed number of iterations after
calibrating the number of steps that we can run in one second. After
that the three trials are run. This should give more stable results.
- Apply more CPU stabilization on Windows
Use SetPriorityClass to apply more CPU priority. Also upgrade
SetThreadPriority to the highest level.
- Always build the Vulkan command buffer test
This catches build regressions more easily. We still skip the test on
non-Android platforms.
Bug: angleproject:2923
Change-Id: I7da234c5af07775ba4a232bb8d65e0138ee7073f
Reviewed-on: https://chromium-review.googlesource.com/c/1330262
Commit-Queue: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Yuly Novikov <ynovikov@chromium.org>
|
|
|
557a1ee4
|
2018-10-30T17:26:24
|
|
Make perf tests faster in correctness-only mode.
When running with "--one-frame-only" we can also skip the test warmup.
Also we can reduce the internal iteration count to 1 to make the tests
as fast as possible.
Bug: angleproject:2923
Change-Id: I2f82ae0dd237767ea7b15074e459ed1094ba9943
Reviewed-on: https://chromium-review.googlesource.com/c/1308737
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Commit-Queue: Jamie Madill <jmadill@chromium.org>
|
|
|
3402d523
|
2018-10-30T15:14:52
|
|
Try to reduce variance in angle_perftests.
This change does a few things:
- make perf test runner script print % variation instead of stddev
This makes it a bit more clear how much variance there is.
- stabilize CPU in the render perf tests
Setting a thread affinity and priority should stop from switching cores
during the run. Hopefully can prevent background noise from changing
the test results.
- warm up the benchmark with a few iterations
This should hopefully make the test results a bit more stable.
- output a new normalized perf result value
The new result is normalized against the number of iterations. So it
should hopefully be stable even if the number of iterations is changed.
- increases the iteration count in the draw call perf tests.
These tests were completely dominated by SwapBuffers time. Increasing
the iterations per step means we actually are bottlenecked on CPU time
instead.
Bug: angleproject:2923
Change-Id: I5ee347cf93df239ac33b83dc5effe4c21e066736
Reviewed-on: https://chromium-review.googlesource.com/c/1303679
Commit-Queue: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Shahbaz Youssefi <syoussefi@chromium.org>
Reviewed-by: Yuly Novikov <ynovikov@chromium.org>
|
|
|
0ea96210
|
2018-10-30T15:14:51
|
|
Vulkan: Enable more perf tests.
Also fixes the check for the correct UINT index extension that wasn't
available on Vulkan. Also includes a workaround for the mock ICD not
implementing buffer state for index ranges.
Bug: angleproject:2923
Change-Id: Iab35809d15f890525a9e658d4148272c46cf1320
Reviewed-on: https://chromium-review.googlesource.com/c/1308733
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Commit-Queue: Jamie Madill <jmadill@chromium.org>
|
|
25224e78
|
2018-10-22T11:56:02
|
|
Vulkan: add GPU trace events
RendererVk now tries, as best as it can, to match the CPU and GPU timers
on init as well as every finish(). A clock-sync event is generated for
each such synchronization point.
RendererVk::traceGpuEvent() is a new function that, given a command
buffer, performs timestamp queries corresponding to GPU events. These
queries are read back when done, without incurring GPU bubbles, at which
point an event is generated with that timestamp.
Bug: angleproject:2908
Change-Id: I08d7d11ff9f8ad6c9f9a9899767c9cd746d0623e
Reviewed-on: https://chromium-review.googlesource.com/c/1296954
Commit-Queue: Shahbaz Youssefi <syoussefi@chromium.org>
Reviewed-by: Yuly Novikov <ynovikov@chromium.org>
|
|
|
4024e217
|
2018-10-17T14:53:29
|
|
perftests: Record trace events to JSON file.
This allows us to view the timeline of events in the trace event
browser. We can extend this to do GPU timestamp queries and analyze
when work is actually in flight.
The trace is enabled in standalone ANGLE only using the flag
--enable-trace with angle_perftests. You can also optionally specify
the trace output file with --trace-file <blah>. The default file is
ANGLETrace.json.
Bug: angleproject:2781
Change-Id: I871f28545d9bf18220b55aaf69e9554dcb4c834d
Reviewed-on: https://chromium-review.googlesource.com/c/1259763
Commit-Queue: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Geoff Lang <geofflang@chromium.org>
|
|
c801e359
|
2017-11-21T13:53:48
|
|
Add configurations of the TexturePerf benchmark that use robust init.
BUG=angleproject:2188
Change-Id: Ia23b54044a85435c761478bceeb9aac961fac0d1
Reviewed-on: https://chromium-review.googlesource.com/783230
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Corentin Wallez <cwallez@chromium.org>
Commit-Queue: Geoff Lang <geofflang@chromium.org>
|
|
|
981f0f8f
|
2017-09-21T09:07:41
|
|
Add flag to do a fast pass through perf tests.
This flag will only render the first frame of each perf test,
regardless of their preferences for how many seconds to run.
It will be useful for speeding up the run time of the perf tests
on testing infrastructure that only cares about correctness.
BUG=chromium:725308
BUG=chromium:765321
Change-Id: I926f488c42f27ef23ef06a0159902613cff04080
Reviewed-on: https://chromium-review.googlesource.com/677306
Reviewed-by: Geoff Lang <geofflang@chromium.org>
Commit-Queue: Jamie Madill <jmadill@chromium.org>
|
|
fca78130
|
2017-09-06T13:51:39
|
|
Extend multiview perf tests to cover all extension code paths
The ANGLE_multiview extension can be supported with three possible code
paths - through view being selected in the vertex shader on D3D and
OpenGL, and through the view being selected in the geometry shader on
D3D only. This patch extends the multi-view performance tests to
benchmark these three different code paths.
BUG=angleproject:2062
TEST=angle_end2end_tests
Change-Id: I443e4db64a95eede1142718a43a095ee5a03738c
Reviewed-on: https://chromium-review.googlesource.com/652466
Commit-Queue: Martin Radev <mradev@nvidia.com>
Reviewed-by: Geoff Lang <geofflang@chromium.org>
Reviewed-by: Olli Etuaho <oetuaho@nvidia.com>
|
|
|
6a6199b4
|
2017-06-05T17:30:55
|
|
Add multiview performance tests
The patch adds two tests to measure the performance of the
ANGLE_multiview extension implementation:
1) The first test renders onto two views by issuing thousands of state
changes and draw calls. The aim of the test is to stress the CPU.
2) The second test renders onto two views by drawing with one draw call
half a million quads with multiple attributes per vertex. The attributes
are passed to the fragment shader where they are used for computing the
color. The aim of the test is to stress the GPU's memory system.
The patch also extends the ANGLEPerfTest's functionality to only run
the benchmark if the necessary extensions are available.
BUG=angleproject:2062
TEST=angle_end2end_tests
Change-Id: Ic63d54a69fdddb72439eeeb1951a500fb1247e95
Reviewed-on: https://chromium-review.googlesource.com/597630
Reviewed-by: Olli Etuaho <oetuaho@nvidia.com>
|
|
|
3764b257
|
2017-02-15T10:41:31
|
|
Vulkan: Run simple triangle perf test.
The most basic perf test possible.
BUG=angleproject:xxxx
Change-Id: I71b28098c0a1f2174a0177b08bddf74d337438e9
Reviewed-on: https://chromium-review.googlesource.com/427270
Reviewed-by: Geoff Lang <geofflang@chromium.org>
Commit-Queue: Jamie Madill <jmadill@chromium.org>
|
|
e1743f37
|
2016-01-06T05:24:22
|
|
Clang warning fixes
This fixes member initialization order and unused private member
warnings that broke the Clang build after the last Angle roll.
BUG=82385
Change-Id: If866f352bda6d66093c180fdbec6c16bf902b628
Reviewed-on: https://chromium-review.googlesource.com/320466
Reviewed-by: Nico Weber <thakis@chromium.org>
Tryjob-Request: Nico Weber <thakis@chromium.org>
Reviewed-by: Geoff Lang <geofflang@chromium.org>
Tryjob-Request: Geoff Lang <geofflang@chromium.org>
Tested-by: Nico Weber <thakis@chromium.org>
|
|
7ee40d1a
|
2016-01-04T08:37:00
|
|
Fix compilation failures on OSX standalone builds
BUG=
Change-Id: Ifd40ae30d601227caf550546fcffc286cf533a73
Reviewed-on: https://chromium-review.googlesource.com/320290
Tryjob-Request: Corentin Wallez <cwallez@chromium.org>
Reviewed-by: Geoff Lang <geofflang@chromium.org>
Tested-by: Corentin Wallez <cwallez@chromium.org>
|
|
8047f065
|
2015-12-15T17:54:42
|
|
Improve perf testing framework
The ANGLERenderTest subclasses have a parameter "iterations". Previously
most of these tests would perform iterations^2 iterations, since the
looping was done both in ANGLERenderTest and in the individual test
classes. Do the looping only in the individual test classes instead.
This enables getting rid of separate beginDrawBenchmark() and
endDrawBenchmark() functions.
Some other unused code is also removed:
1. stepBenchmark function
2. unused parameters to step()
This makes the core loop of running tests simpler.
The perf testing framework also now has shared logic for deciding when
to end a given test.
The score calculation for tests is also changed. Instead of reporting
just the number of operations done, it is reported relative to the actual
run time of the test. This should make the test results more accurate,
since run time of the tests may have some variation. It also enables
changing the run time of the tests without rebaselining them.
In the tests that use GPU, GPU operations are also waited to finish
before stopping the timer.
BUG=angleproject:1261
TEST=angle_perftests
Change-Id: I69e9aad8afd2d9dedd60e144f0a5d4203618feef
Reviewed-on: https://chromium-review.googlesource.com/319381
Tryjob-Request: Olli Etuaho <oetuaho@nvidia.com>
Tested-by: Olli Etuaho <oetuaho@nvidia.com>
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Zhenyao Mo <zmo@chromium.org>
|
|
|
fecbed9f
|
2015-12-17T12:01:12
|
|
Fix code style in perf tests
1) Format switch/case with git cl format
2) Change NULL -> nullptr
3) Change int to GLint where appropriate
4) Use more appropriate variants of asserts
BUG=angleproject:1261
TEST=angle_perftests
Change-Id: I97d6260dbd72ef3897fff145682b7043bf323b76
Reviewed-on: https://chromium-review.googlesource.com/319380
Tryjob-Request: Olli Etuaho <oetuaho@nvidia.com>
Tested-by: Olli Etuaho <oetuaho@nvidia.com>
Reviewed-by: Jamie Madill <jmadill@chromium.org>
|
|
|
5ade8459
|
2015-09-02T11:00:30
|
|
Properly check for EGL_KHR_create_context before creating ES3 context.
Mark EGL_KHR_create_context as supported in the DisplayGLs.
BUG=angleproject:1149
Change-Id: I20671535680eb2c3b9c08205cee243b3aa5c5821
Reviewed-on: https://chromium-review.googlesource.com/297080
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Corentin Wallez <cwallez@chromium.org>
Tested-by: Geoff Lang <geofflang@chromium.org>
|
|
|
f3357ee2
|
2015-07-22T14:10:19
|
|
EGLWindow: remove the unused width and height
EGLWindow does nothing with it per se, but some code was relying on it
to store it. Add width and height to ANGLETest and SampleApplication
instead. Also fix a typo in PerfTestParams, widowWidth -> windowWidth.
BUG=angleproject:1105
Change-Id: I26da607a2e6342864b508a50ee3cf8944608f868
Reviewed-on: https://chromium-review.googlesource.com/287379
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Geoff Lang <geofflang@chromium.org>
Tested-by: Corentin Wallez <cwallez@chromium.org>
|
|
|
9e582b93
|
2015-06-22T11:18:32
|
|
perf_tests: Normalize MS counts in EGL init test.
The un-normalized values depend on the number of frames rendered, so
if we optimize down one time, the other ms values all increase,
because we are rendering more frames. This generates spurious perf
alerts from the perf dashboard.
BUG=None
Change-Id: I1bc480141a8cad7eee96bcdf4d6747a3cbae30f2
Reviewed-on: https://chromium-review.googlesource.com/280980
Reviewed-by: Zhenyao Mo <zmo@chromium.org>
Tested-by: Jamie Madill <jmadill@chromium.org>
|
|
|
dd323e95
|
2015-06-09T15:16:31
|
|
Only run tests that the current hardware can support.
For each config, determine if a context can be created at test
instantiation time. This allows skipping of ES3 tests when the hardware
does not support ES3.
Updated the perf_tests to use the EGLPlatformParameters struct so that
they can be filtered in the same way.
Change-Id: If664604b057cec4005eb4b63bebd83cd4964b7b2
Reviewed-on: https://chromium-review.googlesource.com/276460
Reviewed-by: Corentin Wallez <cwallez@chromium.org>
Tested-by: Geoff Lang <geofflang@chromium.org>
|
|
|
b45a80db
|
2015-05-21T10:39:43
|
|
Make perftests use ANGLE_INSTANTIATE_TEST
This also moves ANGLE_INSTANTIATE_TEST to its own header and makes it
generic over the type of test parameter.
BUG=angleproject:892
Change-Id: Id4e3929d7ad06964b3259015915be84a8ee414f9
Reviewed-on: https://chromium-review.googlesource.com/272553
Reviewed-by: Geoff Lang <geofflang@chromium.org>
Tested-by: Corentin Wallez <cwallez@chromium.org>
|
|
|
3c8870a1
|
2015-04-20T16:00:15
|
|
perf_tests: Use 'score' units.
The dashboard automatically marks 'frames' as 'lower is better', while
'score' is automatically marked as 'higher is better'. Hence, use
score instead of frames.
BUG=468852
Change-Id: I02b3a9e4b74989793d4bfbf21a94e43670b3e028
Reviewed-on: https://chromium-review.googlesource.com/266522
Tested-by: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Geoff Lang <geofflang@chromium.org>
|
|
|
304dcde8
|
2015-04-15T14:26:36
|
|
perftests: Use gtest asserts macro helpers.
We still were using the old cassert code, with a boolean return
value on our initialization code. We can make use of the gtest
macros and helpers to check for a successful init or GL errors.
BUG=476658
Change-Id: I09cbb3d40748cbeaf530ae8f23fb8a1b07e7611f
Reviewed-on: https://chromium-review.googlesource.com/265837
Tested-by: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Geoff Lang <geofflang@chromium.org>
Reviewed-by: Brandon Jones <bajones@chromium.org>
|
|
|
835be965
|
2015-04-13T14:17:49
|
|
perf_tests: Only print final score result.
Don't print all the other measurements and values, since they'll
just add noise to the perf bot dashboards. Also increase the
run time of the DrawCallPerfTest to 10.0 seconds.
BUG=angleproject:744
BUG=468852
Change-Id: If076055ad46b4e6923f39687edcaa25930b188a7
Reviewed-on: https://chromium-review.googlesource.com/265464
Tested-by: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Geoff Lang <geofflang@chromium.org>
|
|
|
514df7f7
|
2015-03-27T09:46:42
|
|
Remove uses of std::unique_ptr in perftests.
This is disallowed in Chromium because it isn't currently portable.
We can put it back later if the rules change.
BUG=angleproject:956
Change-Id: I5be28590f494719b8cc995739dde26726283fced
Reviewed-on: https://chromium-review.googlesource.com/262777
Tested-by: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Brandon Jones <bajones@chromium.org>
|
|
|
7afd4617
|
2015-03-30T10:56:05
|
|
Disable vsync for perf tests and don't report the total time.
BUG=angleproject:955
Change-Id: I8223349622531c8fff7c7cd45bdf4754d7619ae2
Reviewed-on: https://chromium-review.googlesource.com/263062
Tested-by: Geoff Lang <geofflang@chromium.org>
Reviewed-by: Jamie Madill <jmadill@chromium.org>
|
|
|
9bd4dbef
|
2015-03-27T09:46:40
|
|
Split ANGLEPerfTest into a Render and basic test.
The basic test we can use for perf testing internal classes with
mocks. The render tests are more specific to doing draw calls.
BUG=angleproject:956
Change-Id: Iade393facc30c8d7288b1b94a159ce3afe993021
Reviewed-on: https://chromium-review.googlesource.com/262775
Reviewed-by: Olli Etuaho <oetuaho@nvidia.com>
Reviewed-by: Geoff Lang <geofflang@chromium.org>
Tested-by: Jamie Madill <jmadill@chromium.org>
|
|
|
19a43dbe
|
2015-03-20T16:14:04
|
|
Add a perf test for draw call overhead.
These perf tests use the D3D NULL renderers to measure pure CPU
time, with no GPU limitations. Also fix our D3D9 SwapChain code
to skip creating a SwapChain for NULLREF D3D9 devices, and the
vertex decl code to reinitialize the formats when finding a
mismatch with the previous value.
BUG=angleproject:955
Change-Id: I449e63177b48afd1559c36244de0bc252814e813
Reviewed-on: https://chromium-review.googlesource.com/262208
Reviewed-by: Geoff Lang <geofflang@chromium.org>
Tested-by: Jamie Madill <jmadill@chromium.org>
|
|
|
be13ffca
|
2015-03-19T13:42:16
|
|
Use google test to wrap our perf tests.
This provides easy implementation with the Chromium perf bots.
BUG=angleproject:744
Change-Id: I71c04e3d67cae9344ddca9fd95fd3c9849494922
Reviewed-on: https://chromium-review.googlesource.com/260644
Tested-by: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Kenneth Russell <kbr@chromium.org>
|