16 May
2003
16 May
'03
2:40 a.m.
For those of you who don't follow the jack list there was a recent
(semi) discovery htat greatly improves performance for journalled
filesystems.
The main problem is that jack writes data to a /tmp file. It has been
found that mounting /tmp as tmpfs (in RAM) solves the problems that many
people have experienced with lockups while using JACK.
Just put this in your /etc/fstab
none /tmp tmpfs defaults 0 0
I have added this info to the low latency howto also.
--
Patrick Shirkey - Boost Hardware Ltd.
Http://www.boosthardware.com
Http://www.djcj.org - The Linux Audio Users guide
========================================
Being on stage with the band in front of crowds shouting, "Get off! No!
We want normal music!", I think that was more like acting than anything
I've ever done.
Goldie, 8 Nov, 2002
The Scotsman
16 May
16 May
12:50 p.m.
Patrick Shirkey wrote on Fri, 16-May-2003:
For those of you who don't follow the jack list
there was a recent
(semi) discovery htat greatly improves performance for journalled
filesystems.
The main problem is that jack writes data to a /tmp file. It has been
found that mounting /tmp as tmpfs (in RAM) solves the problems that many
people have experienced with lockups while using JACK.
Just put this in your /etc/fstab
none /tmp tmpfs defaults 0 0
This needs to be more fully qualified. These solutions reduce
the audio buffer overruns while using JACK under certain conditions.
It does not solve lockups. The solutions mentioned above (and
below) help when your /tmp directory is mounted on a reiserfs
filesystem. Whether or not it is necessary for other journalled
filesystems is not yet verified.
Note that mounting your /tmp on a tmpfs is not be done without
due consideration. You will need to have ample swap space available
in the case where other applications may write very large files to
the /tmp area.
The other (less intrusive) option to enable JACK to place its
FIFOs on a ram-based filesystem follows:
---------------------------------------------------------------
# mkdir /mnt/ramfs
[edit /etc/fstab and add the following line]
none /mnt/ramfs tmpfs defaults 0 0
Then use --with-default-tmpdir=/mnt/ramfs to the JACK configure
line when you build it. No clients need to be recompiled.
----------------------------------------------------------------
jlc
12:59 p.m.
On Fri, 16 May 2003 13:15:59 +0900
Patrick Shirkey <pshirkey(a)boosthardware.com> wrote:
For those of you who don't follow the jack list
there was a recent
(semi) discovery htat greatly improves performance for journalled
filesystems.
The main problem is that jack writes data to a /tmp file. It has been
found that mounting /tmp as tmpfs (in RAM) solves the problems that many
people have experienced with lockups while using JACK.
I have been plagued with little xruns for a while now, and couldn't
figure out why! I'd seen something on using /tmpfs, but for some odd
reason I thought it was for people using a journalled FS on their root
filesystem. Duh!
I applied this this morning, and have been recording some things, and
have not seen a single xrun during recording!!! Woohoooooo! Hopefully
this puts an end to that little problem.
Thanks, Patrick!
--
======================================================================
Joe Hartley - UNIX/network Consultant - jh(a)brainiac.com
Without deviation from the norm, "progress" is not possible. - FZappa
1:20 p.m.
Joe Hartley wrote on Fri, 16-May-2003:
This is interesting, can you please tell us what your filesystem
configuration is? I am very interested in hearing how this might
help folks with different fs's on /tmp. It was a particularly
dramatic improvement when /tmp was on reiserfs (when I discovered
the problem/solution). If people are seeing xrun problems while
using JACK, I recommend that they try one the options mentioned
in my previous reply email, and let us know the results, and your
fs configuration.
jlc
The main
problem is that jack writes data to a /tmp file. It has been
found that mounting /tmp as tmpfs (in RAM) solves the problems that many
people have experienced with lockups while using JACK.
I have been plagued with little xruns for a while now, and couldn't
figure out why! I'd seen something on using /tmpfs, but for some odd
reason I thought it was for people using a journalled FS on their root
filesystem. Duh!
I applied this this morning, and have been recording some things, and
have not seen a single xrun during recording!!! Woohoooooo! Hopefully
this puts an end to that little problem.
1:29 p.m.
On Fri, 16 May 2003 10:58:42 -0400
Jesse Chappell <jesse(a)essej.net> wrote:
This is interesting, can you please tell us what your
filesystem
configuration is?
Here's my /etc/fstab.
A real DUH moment - I had thought my root fs was ext2, but clearly I made
the switch to ext3 at some point.
The /tape FS is where I do all the ardour recording to.
xtc:~% more /etc/fstab
LABEL=/ / ext3 defaults 1 1
LABEL=/boot /boot ext2 defaults 1 2
none /tmp tmpfs defaults 0 0
none /dev/pts devpts gid=5,mode=620 0 0
none /proc proc defaults 0 0
none /dev/shm tmpfs defaults 0 0
/dev/hdc2 swap swap defaults 0 0
/dev/hdb1 /tape reiserfs defaults,noatime 1 3
I am amazingly happy this morning, since I'd been plagued with little
occasional xruns even though I'd gone through all the low latency and
IRQ stuff. I think this was the final hurdle for me.
Thanks again!!
--
======================================================================
Joe Hartley - UNIX/network Consultant - jh(a)brainiac.com
129 Petta. Lake Rd, Saunderstown, RI 02874 - vox 401.338.9214
Without deviation from the norm, "progress" is not possible. - FZappa
1:36 p.m.
On Fri, May 16, 2003 at 10:48:01AM -0400, Joe Hartley wrote:
I have been plagued with little xruns for a while now,
and couldn't
figure out why! I'd seen something on using /tmpfs, but for some odd
reason I thought it was for people using a journalled FS on their root
filesystem. Duh!
I applied this this morning, and have been recording some things, and
have not seen a single xrun during recording!!! Woohoooooo! Hopefully
this puts an end to that little problem.
NOw that's interesting. What filesystems DO you have?
Do you mean you don't have /tmp on a journalled filesystem, and
this still made a big difference?
--
Paul Winkler
home: http://www.slinkp.com
"Muppet Labs, where the future is made - today!"
1:58 p.m.
On Fri, May 16, 2003 at 10:48:01AM -0400, Joe Hartley wrote:
It was, what fs are you using for /tmp?
Another thing that may help is tuinng the disk elevator with elvtune(8),
the latencies are set pretty high by default, especially for scsi disks
that have thier own block grouping code.
Reducing it will reduce throughput, but may help things get written out
the disk earlier and reduce burstyness.
- Steve
The main
problem is that jack writes data to a /tmp file. It has been
found that mounting /tmp as tmpfs (in RAM) solves the problems that many
people have experienced with lockups while using JACK.
I have been plagued with little xruns for a while now, and couldn't
figure out why! I'd seen something on using /tmpfs, but for some odd
reason I thought it was for people using a journalled FS on their root
filesystem. Duh! 
2:30 p.m.
It was, what fs are you using for /tmp?
Another thing that may help is tuinng the disk elevator with elvtune(8),
the latencies are set pretty high by default, especially for scsi disks
that have thier own block grouping code.
Reducing it will reduce throughput, but may help things get written out
the disk earlier and reduce burstyness.
- Steve
Steve,
I hesitate to take this thread in a different direction, so let me just
ask if this problem is possibly card dependent, or something that has really
been caused by recent changes to Jack?
I have not been able to run Jack now for months, but 3 months ago, when
I was running Jack daily, I had my system on an ext3 drive and audio going
to a reiserfs drive, and with my HDSP 9652 I never once saw an xrun at any
latency setting of 256 or above. I might periodically get one at 128 and
would get a few more at 64. All of that under KDE using RH 8.
Actually, I have run Jack over this period, but only using the on-board
sound chip, so that's pretty useless to me for anything other than looking
at GUI interfaces, but with that driver (snd-via82xx) I did see xruns.
Why did I not see xruns with the HDSP 9652?
Mark
7869
days inactive
7869
days old
linux-audio-user@lists.linuxaudio.org
7 comments
6 participants
participants (6)
-
Jesse Chappell -
Joe Hartley -
Mark Knecht -
Patrick Shirkey -
Paul Winkler -
Steve Harris