Subject: [linux-audio-dev] Measured latency
From: Karl MacMillan (karlmac_AT_peabody.jhu.edu)
Date: Wed Jan 10 2001 - 20:40:06 EET
I have made some end to end audio latency measurements of my system that I
think some of you may be interested in. First, here is the setup:
Redhat 7
2.4.0 w/ Andrew Morton's low latency patch
Alsa latest cvs
Celeron 366 (IBM Thinkpad)
Ess solo1
With this setup I ran a simple thru program that copied data from the line
input of the soundcard, converted the data to floats, converted the data back
to ints, and wrote to the output. To measure the latency I used another
computer to record a stereo track that had the output of the soundcard on one
track and the output of a synthesizer that I was using to create clicks to the
other track. The ouput of the synthesizer also went to the soundcard input,
of course. Here is some ascii art:
Computer B
--------------
--------------------------------------------| L in |
--------- | | |
| synth |---| -------------- | |
--------- |----------| Computer A |-------------------| R in |
-------------- --------------
In this way the difference in start time between the L channel and R channel
recorded with Computer B is the latency introduced by Computer A. All of the
connections were analog so there shouldn't be a measurable delay introduced by
them (at least not measurable by me!).
I was using 3 fragments 64 samples in length (everything was done at 44100hz,
16bit) for a buffer length of 192 samples (or 3.67 ms). The measured latecy
was 539 samples (12.22 ms).
I know that soundcards can potentially add latency in addition to the buffer
size, but my question is whether over 300 samples is the right order of
magnitude for the added latency. It seems a little high to me, but I do not
have enough knowledge of the low-level stuff to know. Any comments?
Thanks,
Karl
_____________________________________________________
| Karl W. MacMillan |
| Computer Music Department |
| Peabody Institute of the Johns Hopkins University |
| karlmac_AT_peabody.jhu.edu |
| www.peabody.jhu.edu/~karlmac |
-----------------------------------------------------
This archive was generated by hypermail 2b28 : Wed Jan 10 2001 - 21:55:17 EET