3fc37deca8
We've been advocating as little usage of streams as possible inside
the system_server, and this is a benchmark showing the costs of using
streams compared to rolling manual logic.
Results on a taimen show that direct route is 2.2x faster (!).
Trial Report (1 of 2):
Experiment {instrument=runtime, benchmarkMethod=timeDirect, vm=app_process, parameters={}}
Results:
runtime(ns): min=47234.28, 1st qu.=47234.28, median=47234.28, mean=47234.28, 3rd qu.=47234.28, max=47234.28
Trial Report (2 of 2):
Experiment {instrument=runtime, benchmarkMethod=timeStreams, vm=app_process, parameters={}}
Results:
runtime(ns): min=105660.39, 1st qu.=105660.39, median=105660.39, mean=105660.39, 3rd qu.=105660.39, max=105660.39
Bug: 77808546
Test: vogar --mode app_process --benchmark frameworks/base/core/tests/benchmarks/src/android/util/StreamsBenchmark.java
Change-Id: Iac43a75970eee2c91593316a1b5a45d2c96c476b
These benchmarks use the Caliper benchmark framework, and can be run on a remote device using Vogar: http://code.google.com/p/caliper/ http://code.google.com/p/vogar/ ------------------------- Quick Command Line Reference: # Build vogar and dependencies. $> mmma -j32 external/vogar # First make sure art has permissions to dalvik-cache, otherwise it will run slower with interpreter. $> adb root # Run vogar in benchmark mode, telling it to use app_process (not dalvikvm which is default) # Otherwise you will likely crash with UnsatisfiedLinkError despite having correct JNI code. $> vogar --mode app_process --benchmark path/to/Benchmark.java # Sometimes your benchmarks might time out, if so increase the timeout: # (--timeout goes to vogar, and --time-limit goes to caliper) $> vogar --timeout 1000 --mode app_process --benchmark path/to/Benchmark -- --time-limit 9999s