|
ab130438
|
2018-11-20T14:09:50
|
|
Skip SetUp and TearDown in VulkanCommandBufferPerfTest if its flaky.
BUG=angleproject:2923
Change-Id: Id201efa09bcc67d4e40154e2362bf4b78ffd5652
Reviewed-on: https://chromium-review.googlesource.com/c/1344591
Reviewed-by: Geoff Lang <geofflang@chromium.org>
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Commit-Queue: Geoff Lang <geofflang@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>
|
|
|
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>
|
|
077cd858
|
2018-08-06T10:46:22
|
|
Vulkan: Introduce Command Buffer Perf Test
Implemented command buffer performance test based on
LunarG samples. Includes a lot of utility functions and
should build/run on Linux, Android
Bug: angleproject:2758
Change-Id: I8606802fb3e8e0621856860c45ef50ba836c26f8
Reviewed-on: https://chromium-review.googlesource.com/1162119
Commit-Queue: Omar El Sheikh <theoking@google.com>
Reviewed-by: Jamie Madill <jmadill@chromium.org>
|