EXS 24 Logic Sampler Users Group

At 10:25 AM 1/18/03 +0100, you wrote: 
>
> <vaz303@...> wrote:
> > Certainly! Forgot to mention that!
> > Was thinking maybe just a 29160 SCSI card and 10000 RPM
> > SCSI HD would do the trick - however these HD's wouldnt be big
> > enough (36 gigs max?) to house a sample library of that size in
> > it.. So yes -  looks like i would need a raid setup :(  $$$
>
> Hm... why $$$?
> As a potential future Mac user I'm really interested in this.
> These days, almost any PC mainboard has some UDMA 133 controller onboard - 1
> or 2 slots, allowing for 2 or 4 total physical drives, in addition to the
> standard IDE ports (which allow for 4 other devices to be connected, such as
> your system disk and CD/CDR/DVD drives).
> A fast and reliable HDD such as, say, a WD 80GB (Maxtors should do the job
> as well) wouldn't cost you more than around 100$ or so, so in the end you
> can get 320GB of fast diskspace for 400$.
> Is that any different on Macs?


Yes -- RAID controllers are not provided on the motherboard. You can of course
buy them from third parties but at additional cost. Like Win2K/XP,  the Mac OS
supports striped volumes in software but the efficiencies gained are not
comparable those of hardware RAID systems -- in fact since all the processing 
is handled by the CPU instead of a hardware controller it loads the CPU --
hardly what you would want on a PC or a Mac DAW. 

>
> And why RAID? On PCs all those controllers allow for a RAID setup, but it
> simply makes no sense - unless you call the standard "Stripe 0" (I think
> that's what the standard is...) a RAID allready. Single drives are fast
> enough to handle tons of audio data simultaneously (on my current mediocre
> Athlon 1GHz I can run 200+ tracks).


Yes even at 13GB drives were fast enough for a sh**load of tracks. The question
is -- how many voices can you get out of ESX24 on your system with streaming on
and just one Audio drive? 

CPU resources could perhaps be the limitation on an Athlon 1GHz system. With a
faster CPU eventually the harddrive would be the limitation. That's where
mirrored drives (for example) are an advantage -- reads are twice as fast. You
should be able to get many more streaming voices from a RAID array where two or
more lots of heads are doing the seeks.  Read access time is the main limit to
be overcome. In continuous reads the latest 7200 rpm drives (eg IBM Deathstar
180GXP) are transferring data to RAM at over 50 MByte/sec which is ~ 300 times
as fast as a CD playing back at 1X (one stereo track at 44.1kHz and 16 bit). 
However the same drive would not be able to support 300 simultaneous stereo
sampler voices because of all the time lost seeking the data. This becomes
relevant with heavily loaded GigaStudio systems where orchestral arrangements
use many simultaneous voices. Now that the big .gig format libraries are
working in EXS24 with streaming the advantages of RAID  come into play.

Regards,
Murray