Silent pc (Re: [ardour-users] Session export ?

[Matthew Sewell]

On Thu, 2005-02-24 at 13:40, Mark Knecht wrote:
> Well, to be clear I didn't write the advice above. That was Gerard.

Yes, I know.  I was just being lazy and killing two birds with one
email.

> He'll have to answer to that. I did do some work looking at this 2-3
> years ago. At the time I would have said from a performance point of
> view that reiserfs was better BECAUSE it somehow places the journal
> along with the files which seemed to cause fewer disk seek operations.
> With the ext3 journal it was a completely separate file which was not
> located near the audio data and seemed to cause disk seeks (and hence
> longer latency) in time with the journaling operation. (THIS WAS A
> GUESS! I watched the operation closely with Benno's latency test
> program and noticed that the misses on latency were the same frequency
> that the journaling was set up for on the drive I did the work on. It
> is NOT a conclusive statement. Only an observation which is now 3
> years old.)
> 
This was also the impression that I got about the journaling.

> I do NOT recommend VFAT even though I am happy and successful using
> it. Personally I have mixed feeling about using journaling at all on a
> strictly audio drive. All forms of journaling by definition are slower
> and take more system resources. If you are worried about losing audio
> data then back it up on a DVD and don't try to make it fault tolerant
> on your system. That's my thought. IT IS NOT A RECOMMENDATION.
> 

Understood.  I back up to DVD anyway.


> 
> My guess would be that something on the machine is getting in the way
> once in awhile and causing these misses. My 1394 drive only does about
> 24MB/S and I get better results. (Or think I do!)
> 
Ah.  I'll bet that, by implementing the changes I mentioned earlier, I
will probably get pretty good performance even with IDE.  I'm not doing
anything spectacular.
 
> > Is that an IDE drive enclosed in a firewire enclosure?
> 
> Yes. I have a bunch of them, but then again I'm a 1394 advocate. 1394
> under Linux is hit and miss. It is pretty important to get the right
> stuff - the right enclosure - the right 1394 adapter, etc. It's very
> possible to get 1394 devices that work fine under Windows and they
> don't function at all under Linux. Once you find the right stuff it's
> great, but for me it was hit and miss.
> 
> >

Hmmm.  I miss is bad when you are on a tight budget.

>  
> 
> The truth is it is not that easy to answer. There are good and bad
> chipset for ALL manufacturers. I am currently using VIa in my desktop
> and ATI in my laptop. Both work fine AFAICT. My GigaStudio machine is
> VIa.
> 
> That said you cannot(probably) buy the exact motherboards I am using
> as they are too old now. You have to buy new VIa chipsets and take
> your chances...
> 
> On the other hand, ANYTHING you have read that is more than 9 months
> old about ANY chipset manufacturer probably relates to a chipset you
> can no longer buy also and there for is not appropriate for a
> decision.

Excellent point.

> 
> (I understand there is language on the Ardour website about this. I
> respect the opinion but I also respectfully disagree. I feel my
> experience is just as valid.
> 

Absolutely!  Learning is about disagreement, isn't it?  (Please don't
agree with me!)