Question 1: The pipe from normal mixer's submix to fast mixer track 0 is needed for devices that can only handle a single output stream in the kernel. This was a problem initially. Nowadays, more devices support mixing in the kernel, so we could remove the pipe. I am filing an internal issue to track this. Please feel free to file an external issue here also, if you want to: https://code.google.com/p/android/issues/entry
You are correct that mixing track 0 from the submix does add computational delay, but it does not add algorithmic delay or significant latency. So removing or disabling the pipe when possible should be more efficient for CPU usage / power consumption, but I don't expect it will decrease latency significantly. Question 2: I am not sure what you mean, by delay caused by SRAM zero data. Do you mean the computational delay caused by writing zeroes into SRAM? If so, that is not a significant contributor to total latency. Or do you mean something else? On Monday, May 11, 2015 at 5:11:53 AM UTC-7, weiwei Zhang wrote: > > I am an android platform developer, recently I am doing some investigation > about audio latency,and i have two question about it , could you help me on > that? > > 1st question: > > I just read the FastMixer code of LP5.0, when play a sound pool stream, a > fast track such as fast track1 will be created and audio flinger will also > write zero data to fast track0 , finally both of them will be mixed in fast > mixer thread. > > My question is why we need audio flinger audio_Out thread write zero data > to fast track0 and then mixed with fast track1,as i known, it will add more > latency for sound pool playback. Whether we can mix fast track 1-7 directly > by fast mixer thread and write them to HAL? > > 2nd question is: > > The definition of warm latency in CDD* is:* > > *"warm output latency*" is defined to be the interval between when an > application requests audio playback and when sound begins playing, when the > audio system has been recently used but is currently idle (that is, silent). > > As I known, when the audio system has been recently used but is currently > idle, audio_out thread in audio flinger will write zero data to SRAM(at > most 3 seconds), when application requests audio playback, should the warm > latency include the delay caused by SRAM zero data? > > > Thank you firstly. > -- -- unsubscribe: android-porting+unsubscr...@googlegroups.com website: http://groups.google.com/group/android-porting --- You received this message because you are subscribed to the Google Groups "android-porting" group. To unsubscribe from this group and stop receiving emails from it, send an email to android-porting+unsubscr...@googlegroups.com. For more options, visit https://groups.google.com/d/optout.