EXS 24 Logic Sampler Users Group

Reading all this talk about the RAID configurations I wonder if 
someone could post a link of the hardware ( RAID + DRIVES ) in there 
system that is running rock solid.

I'm thinking of going:
EXS24 (play back only, no desgin) /VSL + / & external efx processing 
maybe the knew KSP8 with optical option. 

Thanks for past present and future
Jamie

So you want RAID to stream your samples?  Even if you do disk 
streaming, you'll still get a processor load, so your limitation will 
be your CPU not disks.

I'm pretty sure two 10K RPM SCSI drives from seagate striped in RAID 0 
will do for you what you need.

The ATTO UL2S will be sufficient for your need.  If you're using a mac, 
then get the program called
SoftRAID.    Charismac DiskTools or whatever it's called will also work 
OK.  If you're using a PC then
just search for any cheap RAID 0 software.

You probably don't even need two drives and a RAID 0.

Using an external FX processor like the KSP8 is a great idea.  It's 
many times the cost of a UAD or TC card, but it's many times their 
superior in general quality.


On Saturday, March 1, 2003, at 06:42 AM, bitesound2001 
<bitesound@...> wrote:

>
> Reading all this talk about the RAID configurations I wonder if
> someone could post a link of the hardware ( RAID + DRIVES ) in there
> system that is running rock solid.
>
> I'm thinking of going:
> EXS24 (play back only, no desgin) /VSL + / & external efx processing
> maybe the knew KSP8 with optical option.
>
> Thanks for past present and future
> Jamie
>
>
<image.tiff>
>
>
>
>
>
> To unsubscribe from this group, send a blank email to:
>    exs-users-unsubscribe@egroups.com
> For a list of places to get free samples please see:
>    http://groups.yahoo.com/group/exs-users/links/
>
>
>
>
>
> Your use of Yahoo! Groups is subject to the Yahoo! Terms of Service.

[Non-text portions of this message have been removed]

I tend to agree with this for the moment. BUT in Logic 6, this argument
loses steam since you can freeze a ton of tracks thereby converting CPU load
to disk load. A 36-64 track mix with all frozen tracks would definitely
benefit from a RAID, but your right. It's not normally needed if you've got
a 10,000rpm.

> So you want RAID to stream your samples?  Even if you do disk
> streaming, you'll still get a processor load, so your limitation will
> be your CPU not disks.
> 
> I'm pretty sure two 10K RPM SCSI drives from seagate striped in RAID 0
> will do for you what you need.




[Non-text portions of this message have been removed]

At 06:09 PM 3/1/03 -0500, you wrote: 
>
> I tend to agree with this for the moment. BUT in Logic 6, this argument
> loses steam since you can freeze a ton of tracks thereby converting CPU load
> to disk load. A 36-64 track mix with all frozen tracks would definitely
> benefit from a RAID, but your right. It's not normally needed if you've got
> a 10,000rpm.


The other advantage or RAID for big sample libraries is the extra space. SCSI
10K drives are typically pretty modest in size. RAID level 0 arrays (2 or more
drives, no data redundancy) have excellent read performance but are more
vulnerable to data loss than single drives. If one drive fails you lose the
entire array.  So, if you are using them for your own audio files, you really
need to back your data up religiously. 

However, if they are just for streaming samples from sample libaries you have
on DVD or CD the risk is not such a big deal -- if a drive goes down, replace
it and rebuild the array. If your business depends on them buy a spare drive(s)
when you buy the original disks. 

You can also use four drives for a RAID 0 +1 array and have mirroring of all
data (greater security). This would get very expensive with SCSI drives --
remembering the I in RAID stands for "Inexpensive". Personally I would go RAID
O with a hardware RAID controller and some big 60 - 80 GB per platter ATA 7200
or (soon to be available) 10k RPM ATA drives.

A great improvement to the EXS24 would be the facility where you could assign
individual instruments to use the disk streaming feature. I often have mixes
where huge instruments (> 1 GB) are combined with much smaller percussion
instruments (for example) that still use many voices. I could make better use
of my RAM and CPU resources if I could assign the small busy instruments to the
normal "RAM resident" state and let my bigger instruments stream up from the
hard drives. This would save both disk and CPU load because instruments loaded
in 32 bit format don't need to be converted to 32 bit float on the fly but
streaming instruments do. 

Regards,
Murray

From: Slave Audio <slaveaudio@...>

>Even if you do disk
>streaming, you'll still get a processor load, so your limitation will
>be your CPU not disks.

If you're a computer hardware engineer making an informed statement, 
I'll shut up, since I'm not. But I'm going to call your bluff and say 
that while there is some CPU load, the limiting factor is absolutely 
going to be the hard drive.

I can run close to 128 sustained stereo EXS24 voices (that's 256 
voices) on my dual gig Quicksilver G4 using a partition of the 
internal drive.
-- 

Nick Batzdorf
818/905-9101, cell 590-9101, fax 905-5434

Nick Batzdorf wrote: 
>
> From: Slave Audio <slaveaudio@...>
>
> >Even if you do disk
> >streaming, you'll still get a processor load, so your limitation will
> >be your CPU not disks.
>
> If you're a computer hardware engineer making an informed statement, 
> I'll shut up, since I'm not. But I'm going to call your bluff and say 
> that while there is some CPU load, the limiting factor is absolutely 
> going to be the hard drive.
>
> I can run close to 128 sustained stereo EXS24 voices (that's 256 
> voices) on my dual gig Quicksilver G4 using a partition of the 
> internal drive.


Hi Nick,

That is a lot of voices from an ATA drive. Can you tell us what model drive is
it and how is it formatted? ( cluster size)

I am not sure what point it serves to make though. I take it you are not at the
limit of CPU with one drive. So, what is to stop you spreading your samples
around? 

Here are four options that could be used simultaneously if you you wanted to go
the whole hog.
1) do as you have done and put a bunch of samples on a partition of your system
disk,  
2) another lot on a SCSI sub-system (RAID for max performance)
3) some on your Audio drive on the second ATA channel. 
4)You could add more on a Firewire drive too.  

Eventually CPU or PCI bandwidth will be the limit.

Regards,
M

This is not correct. Streaming is limited by the disk speed and maximum
bandwidth of throughput. Of course there's some minor CPU load but it's
negligible if no plugins are running (which is the case with freeze since
freeze creates a bounced file and disables the plugins). I completely
disagree that CPU is your limitation here. I know many people who have
proven this fact in their studio work.

>> >Even if you do disk
>> >streaming, you'll still get a processor load, so your limitation will
>> >be your CPU not disks.




[Non-text portions of this message have been removed]

You're not the first person to tell me this.  Maybe I was doing 
something wierd.

I know for sure though, that you can't get 128 tracks streaming off of 
one drive
with multiple samples and kits.  One kit 128 tracks is easy, but with 
multiple
samples the disk read arm flies all over the place and won't be able to 
keep up.


On Sunday, March 2, 2003, at 01:21 PM, Jer Olsen wrote:

> This is not correct. Streaming is limited by the disk speed and maximum
> bandwidth of throughput. Of course there's some minor CPU load but it's
> negligible if no plugins are running (which is the case with freeze 
> since
> freeze creates a bounced file and disables the plugins). I completely
> disagree that CPU is your limitation here. I know many people who have
> proven this fact in their studio work.
>
> >> >Even if you do disk
> >> >streaming, you'll still get a processor load, so your limitation 
> will
> >> >be your CPU not disks.
>
>
>
>
> [Non-text portions of this message have been removed]
>
>
<image.tiff>
>
>
>
>
>
> To unsubscribe from this group, send a blank email to:
>    exs-users-unsubscribe@egroups.com
> For a list of places to get free samples please see:
>    http://groups.yahoo.com/group/exs-users/links/
>
>
>
>
>
> Your use of Yahoo! Groups is subject to the Yahoo! Terms of Service.

[Non-text portions of this message have been removed]

Nick Batzdorf wrote:

>  > I can run close to 128 sustained stereo EXS24 voices (that's 256
>>  voices) on my dual gig Quicksilver G4 using a partition of the
>  > internal drive.
>

From: Murray McDowall <murraymc@...>

>That is a lot of voices from an ATA drive. Can you tell us what model drive is
>it and how is it formatted? ( cluster size)

It's just the 80GB internal drive that comes with the Mac 
(partitioned into 6GB + all the rest). But I'm not sure those voices 
are all streaming - in fact I suspect they're all playing from memory.

>I am not sure what point it serves to make though. I take it you are 
>not at the
>limit of CPU with one drive. So, what is to stop you spreading your samples
>around?

I'm not serving a point, I'm just putting samples in the most 
convenient place. It's hard to think of anything else useful for that 
much storage space!

>Here are four options that could be used simultaneously if you you 
>wanted to go
>the whole hog.
>1) do as you have done and put a bunch of samples on a partition of 
>your system
>disk,\ufffd
>2) another lot on a SCSI sub-system (RAID for max performance)
>3) some on your Audio drive on the second ATA channel.
>4)You could add more on a Firewire drive too.\ufffd
>
>Eventually CPU or PCI bandwidth will be the limit.

Nobody talks about the other limit, which is 64 voices per EXS24. 
That will come into play very quickly with the new streaming 
libraries. People are filling up all 160 Gigastudio voices with one 
complicated program. Remember, an 8-layer crossfaded program uses 
eight channels for each voice.
-- 

Nick Batzdorf
818/905-9101, cell 590-9101, fax 905-5434

>Nick Batzdorf wrote:
>
>
>Nobody talks about the other limit, which is 64 voices per EXS24.
>That will come into play very quickly with the new streaming
>libraries. People are filling up all 160 Gigastudio voices with one
>complicated program. Remember, an 8-layer crossfaded program uses
>eight channels for each voice.
>--
>
>Nick Batzdorf
>818/905-9101, cell 590-9101, fax 905-5434

Very true.  With my EMU samplers and the Siedlazcek library, string 
crossfade patches that would play fine on my 128 voice E4 would voice rob 
on my 64 voice E5000.  Definitely something to keep in mind when porting 
Giga or any other format to EXS24.

Sean McCoy
Oregon Sound Recording