[alsa-devel] buffer time
What's the correct buffer time to use with alsa?
since I have following in ags.sf.net
buffer size: 940 alsa iterations: 47 per second
Since I don't have segmentation of tact in my application BPM is the only scale I got really slow output. Further can't run anymore with realtime priorities. I would prefer using 940 because its near a good throughput rate. Years ago I calculated it the exact value would be something 93.whatever
Please help
best regards Joël
Am Montag, den 18.08.2014, 22:34 +0200 schrieb Joël Krähemann:
What's the correct buffer time to use with alsa?
since I have following in ags.sf.net
buffer size: 940 alsa iterations: 47 per second
Since I don't have segmentation of tact in my application BPM is the only scale I got really slow output. Further can't run anymore with realtime priorities. I would prefer using 940 because its near a good throughput rate. Years ago I calculated it the exact value would be something 93.whatever
Please help
best regards Joël
Sorry there went something wrong it should be 943.dontknow
On 08/18/2014 10:34 PM, Joël Krähemann wrote:
What's the correct buffer time to use with alsa?
Playback or capture?
Anyway, Linux performs at its very worst when faced with typical audio/video work, because it tends to postpone things and execute them in a single big chunk. This is good for throughput, as it can re-order everything so that less harddisk head seeks are required, but it causes an application that, say, writes out data at a constant but quite low rate, to experience latencies of well over 2 seconds on an otherwise idle system.
Because of that, my answer to such a question it usually: All the memory that you can spare. In case of ALSA, that means get the biggest you can possibly get, because the DMA buffer is usually pre-allocated when the driver probed, so it's going to waste if you don't use it.
Am Montag, den 18.08.2014, 22:34 +0200 schrieb Joël Krähemann:
What's the correct buffer time to use with alsa?
since I have following in ags.sf.net
buffer size: 940 alsa iterations: 47 per second
Since I don't have segmentation of tact in my application BPM is the only scale I got really slow output. Further can't run anymore with realtime priorities. I would prefer using 940 because its near a good throughput rate. Years ago I calculated it the exact value would be something 93.whatever
Please help
best regards Joël
I've just found
https://perf.wiki.kernel.org/index.php/Tutorial#multiplexing_and_scaling_eve...
thank you Joël Krähemann
participants (3)
-
Clemens Ladisch
-
Joël Krähemann
-
Mike Looijmans