[linux-audio-user] The trouble with disks, part II

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

Subject: [linux-audio-user] The trouble with disks, part II
From: John Anderson (ardour_AT_semiosix.com)
Date: Tue Mar 23 2004 - 20:06:16 EET


Or should that be part III? I can't remember. Whatever. Let the saga
continue.

Several months ago, I posted here about my xrun problem. A brief
refresher: MSI K7N2Delta mobo, Athlon XP2200+, 1Gb RAM, Adaptec 29160,
Fujitsu 36Gb (MAN3367MP), Terratec EWS88MT, Matrox G550,
linux-2.4.18-22, 2.6.[0-4]. With the invaluable help of various denizens
of this list, I went through the list of possibilities from irq
priorities to kernel patches to jack compile flags. Eventually somebody
found an article about problems with the Adaptec 2940, and suggested I
try an IDE drive. Which was about the only thing I hadn't done.

In the meantime I've moved house, cut my hair, set a date for my wedding
and bought a car. So it being a time for major life changes, today I
went and bought an IDE drive - Seagate Barracuda 40Gb (ST340014A).
Ordinary off-the-shelf nothing special. Not even serial-ata. Walked into
a shop and said "Hello, I need to buy a 40Gb IDE drive". Walked out 10
minutes later.

Previously I couldn't even run at -n 2 -p 1024 without a glitch every
few seconds, unless I preloaded the .wav files for the ardour session to
get them in cache.

With the audio data (reiserfs) and the swap partition on the IDE drive
I'm running jack at -n 2 -p 128 with ardour and 6-10 tracks and not an
xrun in sight. I can play an *entire* session and not one xrun shows up.
Woohoo! linux-2.6.3-mm2 FWIW, although jack got unhappy and kicked
ardour off the graph when I added jamin and bounced the cpu to 75%. Not
entirely unexpectedly.

What burns is that I bought the SCSI setup specifically for audio.
Could've saved quite a bit of money. Ah well. Maybe I can sell it or
something. I spose I could try to get hold of a non-Adaptec SCSI
controller, but it seems like more hassle than it's worth.

Oh, and some hard performance numbers, disc0 is SCSI, disc1 is IDE:

bash-2.05b# hdparm -Tt /dev/discs/disc0/disc

/dev/discs/disc0/disc:
  Timing buffer-cache reads: 128 MB in 0.34 seconds =375.42 MB/sec
  Timing buffered disk reads: 64 MB in 1.31 seconds = 48.75 MB/sec

bash-2.05b# hdparm -Tt /dev/discs/disc1/disc

/dev/discs/disc1/disc:
  Timing buffer-cache reads: 128 MB in 0.35 seconds =364.73 MB/sec
  Timing buffered disk reads: 64 MB in 1.14 seconds = 56.05 MB/sec

bye
John


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

This archive was generated by hypermail 2b28 : Tue Mar 23 2004 - 20:11:33 EET