Hello
I'm running an a Pd patch that reads and writes to disk simultaneously with a number readsf~ and writesf~ objects. It runs very well using alsa or with oss emulation but produces occasional clicks when running with jack. I'm wondering if there's anything I can do to fix this?
Latency is not an issue with this application and i'm running the following to set up jack:
/usr/bin/jackd -R -dalsa -dhw:0 -r44100 -p2048 -n2 &
Pd includes the following start options: -rt -jack
Have compiled pd-0.38-4 with the following configuration: ./configure --enable-newhash --enable-jack --enable-fftw --enable-simd --enable-optimize=pentium4 --enable-threadedsf --enable-atomic – enable-debug
I'm using a Gentoo 2.6.10 kernel. Also using realtime-lsm to get non-root realtime priority (althought this feature doesn't seem to create any audible benefits or problems in this particular patch). The computer is a Compaq Presario R3000 laptop using atiixp alsa drivers.
alsa is being loaded as a module and not compiled into the kernel (haven't tried compiling into the kernel)
When clicks occur, Pd reports errors such as: 8.41 A/D/A sync 18.41 A/D/A sync 18.41 A/D/A sync 18.78 unknown 18.78 unknown 18.78 unknown
Can anyone please suggest what might be causing the glitches? (Ardour by-the-way seems to run well heavily loaded, without any glitches)
Best, Iain
Iain Mott www.reverberant.com
Have compiled pd-0.38-4 with the following configuration: ./configure --enable-newhash --enable-jack --enable-fftw --enable-simd --enable-optimize=pentium4 --enable-threadedsf --enable-atomic â enable-debug
hm ... are you running 0.38-4 or devel_0_38? these specific flags only make sense for devel ... if you're on devel, please make sure too check out the latest sources and tweak the sleepgrain values ...
cheers ... tim
Thanks Tim and Pat
No i wasn't running devel. I've tried what you suggest but unfortunately it still clicks. At least now there is more detailed debugging info when it glitches, usually something like:
subgraph starting at pure_data_0 timed out (subgraph_wait_fd=11, status = 0, state = Running) **** alsa_pcm: xrun of at least 1.940 msecs **** alsa_pcm: xrun of at least 0.007 msecs jack most likely kicked us out ... trying to reconnect
I recompiled omitting various flags in my original configuration line, but to no effect. Tried starting Pd with variety of sleepgrain sizes from 0.01 to 300 (for the hell of it). Used the -cb_scheduler startup flag too. I noticed with this, it adds a "cb_scheduler" checkbox in the media-->Audio Settings dialogue which is unchecked. Unfortunately checking it (and applying) doesn't fix it.
Just in case I somehow botched the install here's what I did (on the first attempt at least):
cvs -z3 -d:pserver:anonymous@cvs.sourceforge.net:/cvsroot/pure-data co -r devel_0_38 pd cd pd ./bootstrap.sh ./configure --enable-newhash --enable-jack --enable-fftw --enable-simd --enable-optimize=pentium4 --enable-threadedsf --enable-atomic --enable-debug make su - cd /home/iain/pd make install
Ran pd with: -rt -jack -sleepgrain 0.1 -cb_scheduler
When I start Pd it identifies as: Pd version 0.38.4 devel {OSS 2} {ALSA 1} {jack 5} courier
the version of jackd is 0.99.0 and alsa driver is 1.0.8
Any help greatly appreciated.
cheers, iain
On Sun, 2005-05-01 at 09:30 +0200, Tim Blechmann wrote:
Have compiled pd-0.38-4 with the following configuration: ./configure --enable-newhash --enable-jack --enable-fftw --enable-simd --enable-optimize=pentium4 --enable-threadedsf --enable-atomic – enable-debug
hm ... are you running 0.38-4 or devel_0_38? these specific flags only make sense for devel ... if you're on devel, please make sure too check out the latest sources and tweak the sleepgrain values ...
cheers ... tim
the version of jackd is 0.99.0 and alsa driver is 1.0.8
hm ... just a guess, but alsa 1.0.8 is reported to be broken ...
cheers ... tim
if you are running devel, try this flag: -cb_scheduler. sleepgrain should be 0.1 or something like that.
please tell us if it's working. pat
----- Original Message ----- From: "Iain Mott" mott@reverberant.com To: pd-list@iem.at Sent: Sunday, May 01, 2005 2:37 AM Subject: [PD] jack clicks but alsa or oss runs smoothly
Hello
I'm running an a Pd patch that reads and writes to disk simultaneously with a number readsf~ and writesf~ objects. It runs very well using alsa or with oss emulation but produces occasional clicks when running with jack. I'm wondering if there's anything I can do to fix this?
Latency is not an issue with this application and i'm running the following to set up jack:
/usr/bin/jackd -R -dalsa -dhw:0 -r44100 -p2048 -n2 &
Pd includes the following start options: -rt -jack
Have compiled pd-0.38-4 with the following configuration: ./configure --enable-newhash --enable-jack --enable-fftw --enable-simd --enable-optimize=pentium4 --enable-threadedsf --enable-atomic – enable-debug
I'm using a Gentoo 2.6.10 kernel. Also using realtime-lsm to get non-root realtime priority (althought this feature doesn't seem to create any audible benefits or problems in this particular patch). The computer is a Compaq Presario R3000 laptop using atiixp alsa drivers.
alsa is being loaded as a module and not compiled into the kernel (haven't tried compiling into the kernel)
When clicks occur, Pd reports errors such as: 8.41 A/D/A sync 18.41 A/D/A sync 18.41 A/D/A sync 18.78 unknown 18.78 unknown 18.78 unknown
Can anyone please suggest what might be causing the glitches? (Ardour by-the-way seems to run well heavily loaded, without any glitches)
Best, Iain
Iain Mott www.reverberant.com
PD-list@iem.at mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/listinfo/pd-list
hello - just reporting back on the partial resolution of a problem I was having with audio clicking when running jack (still getting the odd click - but it's good on the whole). Turned out the problem was to do with the delay time applied after sending an "open" message to readsf~. Initially I had delays of 10 msec set before sending "1" to start reading. Turned out a delay of 100 msecs is needed to avoid the clicks.
Without running Jack there is no need for me to apply any delay - no clicks.
Is there a reason for this?
cheers, iain
On Sun, 2005-05-01 at 16:37 +1000, Iain Mott wrote:
Hello
I'm running an a Pd patch that reads and writes to disk simultaneously with a number readsf~ and writesf~ objects. It runs very well using alsa or with oss emulation but produces occasional clicks when running with jack. I'm wondering if there's anything I can do to fix this?
-
Iain Mott -
patrick -
Tim Blechmann