Skip to contentSkip to conversation

: Very high latency with MIDI soft + dmix + ALSA PCM

ALSA - lib 0000765 [email protected] 1 participant 1 message Jul 8, 2005

lore
Share & tools
lore
#0 lore raw

The following issue has been CLOSED
======================================================================
<https://bugtrack.alsa-project.org/alsa-bug/view.php?id=765>
======================================================================
Reported By: ccasteyde
Assigned To:
======================================================================

Project:                    ALSA - lib
Issue ID:                   765
Category:                   0_general
Reproducibility:            always
Severity:                   major
Priority:                   normal
Status:                     closed
Resolution:                 fixed
Fixed in Version:           
======================================================================
Date Submitted:             12-31-2004 13:54 CET
Last Modified:              07-08-2005 10:34 CEST

======================================================================
Summary: Very high latency with MIDI soft + dmix + ALSA PCM
Description:
There is a high latency problem with the ALSA MIDI sequencer and the PCM interface, and this is even more true when dmix is used.

My goal was to use Timidity or fluidsynth as soft wavetable to emulate MIDI port 128:0 and mix their output with other PCM applications through dmix. It's nearly impossible, since in this configuration sound is hashed as if the synth software was unable to output data while dealing with each MIDI tune. I think nearly 30% sound is lost during this process.

The funny thing is that forcing Timidity to use aRts as output, and configuring aRts itself to use ALSA pcm dmixer makes everything works nicely. I think this is due to the extra desynchronisation and buffering due to the main aRts thread, which lets Timidity release ALSA resource before sending the output to the DMA card.

I therefore suspect a locking problem/resource contention in ALSA driver/ALSALib. If this proves true, and cannot be fixed with fine locking rules, I think it is in ALSA lib the problem should be fixed rather than in each sound software (the problem appears with both Timidity and Fluidsynth and certainly all software that use both input and ouput ALSA devices).
======================================================================

----------------------------------------------------------------------
jdthood - 03-19-05 09:22
----------------------------------------------------------------------
Submitter: Has there been any progress on this issue?

----------------------------------------------------------------------
ccasteyde - 03-19-05 14:39
----------------------------------------------------------------------
Well, I tried to play with ALSA midi and sequencer input and output with Timidity (same version) on an updated slack 10.1 + 2.6.11 kernel (vanilla, rebuilt) + NPTL 2.3.4 (rebuilt) + alsalib 2.6.8 (from slack), and I didn't managed to reproduce with and without dmixing. There are still some problems when I do other tasks on the computer, but at least it doesn't crackle anymore if alone now.

Don't know if it was kernel, sound driver, or alsalib that caused that, but it seems to be fixed.

Issue History
Date Modified Username Field Change
======================================================================

12-31-04 13:54 ccasteyde      New Issue                                    
12-31-04 13:54 ccasteyde      File Added: asound.conf                      
03-19-05 09:22 jdthood        Note Added: 0004014                          
03-19-05 14:39 ccasteyde      Note Added: 0004046                          
03-19-05 14:44 perex          Status                   new => resolved     
03-19-05 14:44 perex          Resolution               open => fixed       
03-19-05 14:44 perex          Assigned To               => perex           
03-19-05 14:44 perex          Assigned To              perex =>            
07-08-05 10:34 perex          Status                   resolved => closed  

======================================================================

-------------------------------------------------------
This SF.Net email is sponsored by the 'Do More With Dual!' webinar happening July 14 at 8am PDT/11am EDT. We invite you to explore the latest in dual core and dual graphics technology at this free one hour event hosted by HP, AMD, and NVIDIA. To register visit http://www.hp.com/go/dualwebinar