Re: [ecasound] Glitching

New Message Reply About this list Date view Thread view Subject view Author view Other groups

Subject: Re: [ecasound] Glitching
From: Kai Vehmanen (k_AT_eca.cx)
Date: Mon Sep 24 2001 - 01:02:20 EEST


On Sun, 23 Sep 2001, William Goldsmith wrote:

> processing. I just did a kernel & oss driver upgrade & I upgraded ecasound
> to 2.0.2 using the binary rpm. It's a more or less stock RH 7.1 server
> install.
[...]
> I had a lot of problems with glitches during playback during cpu-usage
> bursts and intermittent clicks & pops as files start or end. I reverted
> back to 1.9dev5 (also via binary rpm) & the problem went away.

I'd bet my money on -z:intbuf ... ie. add -z:intbuf to the 2.0.2
command-line. With 1.9dev5 you should get 2.0.2-ish behaviour by giving
-z:nointbuf .. -z:intbuf was made the default in 1.9dev6. It helps in
certain realtime uses, but performs badly on others (for instance simple
playback or recording). Hope this helps..

> I'm curious as to whether anyone has any ideas about why this would be
> happening. I'm using SBLive cards, IDE drives (tweaked for speed - they test
> at about 20 Mb/s), & a variety of processors: PIII 900 mHz, Athlon 850 & PII
> 450. All systems have 256 mB RAM

The potential problem areas (which don't necessarily overlap) are:
        - scheduling latency (possible caused by kernel drivers,
          kernel itself, system load)
        - disk i/o latency (kernel buffering subsystem, harddisk
          interface, amount of disk traffic, etc)
        - CPU-performance (quite a rare problem actually, nothing but
          getting a faster computer helps here... or using fewer
          effects, etc)
        - disk i/o performance (with todays disks, this is also quite
          rare if you are just streaming a few channels)

Unfortunately the first two problems are most common, and to make it
worse, they usually cannot be solved by getting bigger&better hardware.
-z:intbuf/nointbuf is related to scheduling latency. With -z:intbut you
sacrifice some latency to gain reliability (longer buffer between
soundcard's current playing position and ecasound's current write pos).

Hmm, I guess it's best if -z:intbuf is made the default once again in
2.0.3. The 2.1.x series will provide a better mechanism (edi-15), but
it'll take some time.

> www.radioparadise.com

Btw; A great station! I've been a regular listener for quite some
     time now. Keep on the good work!

-- 
 http://www.eca.cx
 Audio software for Linux!

-- To unsubscribe send message 'unsubscribe' in the body of the message to <ecasound-list-request_AT_wakkanet.fi>.


New Message Reply About this list Date view Thread view Subject view Author view Other groups

This archive was generated by hypermail 2b28 : Mon Sep 24 2001 - 01:01:45 EEST