Assigned
Status Update
Comments
cc...@google.com <cc...@google.com> #2
Update: according to the trace, dex2oat starts at the same moment as the application. So i guess, the problem lays in xiaomi firmware that starts the optimization with the application during benchmark
ca...@google.com <ca...@google.com> #3
Sorry for the delay! I'll admit I'm a little confused about which exact section is what we want to be recording here. Could we potentially use a trace based metric for one of the Animators?
In any event, I'm not sure of any existing, generalized prior art for this. The strategies I know about use manually dropped tracepoints (which could potentially be an option here?). For your solution, how are you thinking we'd know what was an "easy part"?
Description
Component used: Version used: Devices/Android versions reproduced on:
If this is a bug in the library, we would appreciate if you could attach:
I want to use macrobenchmark to analyse launch animation of launcher.
However, launch animation is different with ordinary application animation.
take OPPO phone's launcher as a example( the first screenshot)
the launching animation has three layers :
starting window layer, which cover on the launched app layer be launched app layer, which will be shown after app launched and get screen shot of this app the launcher layer as background, which is blured and disappered
In the perfetto trace , we can see three actual frametimeline for three layers.( the second screenshot)
by sql query, we can see three actual frametimeline slices with same name's layer_name(the third screenshot)
we can't measure the actual frame timeline of launcher, because it's just the launcher background
how do you do in this situation(how to measure the launch aniamtion)?