Age | Commit message (Collapse) | Author |
|
This will reduce build warnings.
Bug: 329456999
Test: mmm external/oboe
Test: mmma cts/apps/CtsVerifier
Test: mmm frameworks/av/media/libaaudio
Change-Id: I5c1e62db88f92af607851c86172d23e30a0d4daa
|
|
|
|
|
|
|
|
Set phase calculation as invalid when magnitude is low.
Some tests has phaseJitter=0.
Also some tests had very high phaseJitter and
should have failed. So I lower the max allowable jitter.
|
|
|
|
8be3b19e26
Original change: https://android-review.googlesource.com/c/platform/external/oboe/+/2982840
Change-Id: I6c4499fff1f97beeb3213c214c789cd98bfd8aec
Signed-off-by: Automerger Merge Worker <android-build-automerger-merge-worker@system.gserviceaccount.com>
|
|
Merging latest Oboe into external/oboe.
Bug: 326989822
Test: mmm external/oboe
Test: mmma cts/apps/CtsVerifier
Test: mmm frameworks/av/media/libaaudio
Change-Id: Id89715e13fb36e5bb116547b42fe725249f49e15
|
|
|
|
Improved DataPaths and Glitch tests
|
|
Describe new data_paths test parameters
|
|
Glitch test now allows the user to select the input and output channels for analysis.
Data Paths now selects the highest priority device type
that is available and then chooses an input device type to match.
This will prevent playing a signal on USB and listening on the mic.
Display types being tested.
Remove some dead code.
Use levels None, Some, All for controlling channel mask testing.
Show up to 8 VU meters in Test Input
|
|
|
|
If policy is Unspecified then return isMMapSupported.
Fixes #1967
|
|
|
|
|
|
* Listen to USB device de/attach event.
With listening to USB device de/attach event, disconnect
test can well sync on the peripherals plug state.
|
|
There are often glitches counted in the manual Glitch test that are not visible or audible!
Those "glitches" were caused by slowly drifting sample rates.
Apparently input and output may be on different clocks!
The fix was to continuously adjust the phase of the reference sine wave
so that it tracks the incoming signal. This is like a "phase locked loop".
Also:
* Improve display of glitches, add cursor
* Add "Auto draw" checkbox
* Add "Force glitch" checkbox
* Replace abs() with fabs() calls.
* Bump OboeTester version to 2.5.9
|
|
|
|
Call an internal method that does not print the warning.
Fixes #1949
|
|
Bump to Oboe V1.8.1
|
|
Was using SL_ANDROID_STREAM_NOTIFICATION
should be SL_ANDROID_STREAM_RING
Fixes #1945
|
|
|
|
Add link to OboeTester_DynamicCpuLoad
|
|
The latency analysis was too slow on lower end devices.
This use a rough pass to find an approximate latency.
Then it does a fine grained analysis in a small window
around the approximate peak to find the exact value.
Fixes #1939
|
|
This will make it easier to dial in a specific workload
and also is more robust.
|
|
|
|
Fixes #1053
|
|
Fixes #403
|
|
|
|
|
|
|
|
|
|
|
|
|
|
Fixes #1928
|
|
|
|
|
|
|
|
|
|
|
|
Call getCommunicationsDevice() periodically and show the result.
This can be used to detect the "grace period" after calling
setCommunicationDevice().
Also cleaned up NativeSniffer.
|
|
|
|
* External Tap-to-Tone
* Tap-to-Tone
* Auto Glitch
|
|
Fixes #1906
|
|
Reduce tolerance for glitches.
Ask user to plugin loopback adapter and turn up volume.
Print peak.amplitude and sine.amplitude in report.
Warn if peak.amplitude too low.
Don't compare with other passed tests if none passed.
Fix SNR calculation. Was always 0.0.
Fixes #1902
|
|
Fixes #1904
|
|
Show CPU migrations in a text display. "-1--45-7"
Fix clearing of CPU affinity.
Disable drawing during the benchmarking phase so it will converge.
|
|
A new TestColdStartLatency.cpp file was added.
|
|
|