@cm said:

instead of trying to have all ca. 400.000 samples on your fingertips it would be more efficient to load selected articulations (eg. starting with basic all) and go into details while working on your piece.
also i would choose more disks holding 120-160GB than a single one with 200 or even more GB to reduce the latency while accessing too much files from a single disk - this would be definately the bottleneck working with a large number of tracks
christian

Would formatting two 160 GB drives in a RAID-0 (striped) array give you even better performance than formatting them separately and storing different sections on each?

If you had vsl in your hands and didn't have a computer. What kind of computer would you have bought (for opus 1 use?)

Since you have more knowledge than me...

[quote=Audun Jemtland] What kind of computer would you have bought (for opus 1 use?)

I guarantee I have less knowledge than you about computers Audun, but based on Antons demos, I would go for an Apple G5

valerie, all tests i did showed that striping (raid 0) speeds up access significantly, the best result so far are noticed with 10.000 rpm SATA drives.
for those not familiar with that: striping means access to 2 disks simultaneously, so while seek-time is the same, data throughput is almost doubled (you should use two identical models of disks for best results)
one thing that doesn't work at all for streaming is to use the type of onboard raid controllers from promise (and others) for creating a raid, whereas the much more expensive (SCSI-)models from mylex,adaptec and others work fine. W2K and XP can create software raids which is cheap and performant, but please note that raid 0 doubles the risk of loosing all data from the striped disks, because if one disk fails, you're lost
christian

Are there any 10K RPM SATA drives other than the Western Digital Raptors? The problem with them, of course, is that the largest ones are "only" 72GB.

nick, have not found any so far - i'd assume we'll have to wait a while until we see the first 146 GB disks with 10.000 rpm - but to have four of them would give you a nice amount of disk space [;)]
christian

btw: has somebody tried the built-in raid option with panther?

If you want to go larger and/or faster than the WD Raptor these days you still have to go Ultra2-SCSI. Quite expensive... [H]

PolarBear

Thanks guys this was very very useful. Also good to hear from you again Nick!

@Another User said:

I'd assume we'll have to wait a while until we see the first 146 GB disks with 10.000 rpm

My understanding is that disks of all sizes cost the same to make. The difference is that the yield is lower for larger ones. (Which of course negates my first sentence, but the point remains!)

nick, i'm sure they have the same thermal problems as the IBM (=hitachi) 146GB - i don't even risk to use a 72 GB raptor without an additional fan
christian

Hm. I hadn't thought about the heat. And I have a 36 and 72 Raptor in one of my machines...I'd better check the cooling!

@Nick Batzdorf said:

Hm. I hadn't thought about the heat. And I have a 36 and 72 Raptor in one of my machines...I'd better check the cooling!

Nick and others with 10,000 RPM drives,
Please give us a report on the poly and performance you get when you migrate to 3.0 (Orchestra) . Many of us are anxious to see what kind of performance you get.

I'll try, but the machine with the Raptors doesn't have Giga on it. As of now, Giga runs on relatively cheap PCs, so that would be overkill!

Dennis,

We have a series of free Street Smart Guides that answer your questions:
http://www.truespec.com/downloads/index.cfm

In general with the Pro Edition, I would plan for two systems and then wait for the release of Giga 3.0 since the increased polyphony will make two today work like four this spring.

Normally, we put strings, percussion and harp on one system and woodwinds and brass on the other.

Peter Alexander
www.truespec.com
310-559-3779

@Nick Batzdorf said:

Are there any 10K RPM SATA drives other than the Western Digital Raptors? The problem with them, of course, is that the largest ones are "only" 72GB.

No, these are the only ones. And as I've said elsewhere, we need to look beyond the statement of "unlimited polyphony" since the Giga brochure clearly stated that with a 2.8GHz system, they slightly doubled the existing polyphony. And this was using two IDE drives in a RAID.

Raptor drives are within a few dollars of their SCSI cousins. The 74GB drive, my cost as a dealer, is roughly $250. Quite a few motherboards come with a SATA RAID, but this is only for TWO drives. So with Vienna, you either have to justify spending $500+ to warehouse just under 150GB of data or spend $340 thereabouts to wherehouse just under 320GB using two 160GB drives. But if you want comparable to the 320GB with Raptors, then the cost is $1000 for hard drives only plus approximately $100 for the RAID card, which you'll set for 0 to have the warehousing space needed for Vienna.

At the risk of offending Tascam, I have to urge caution.

Unlimited polyphony is not really true.

The software in absence of hardware is capable of unlimited polyphony (just like how programs used to advertise themselves as being able to record unlimited audio tracks). But this is really a misnomer since software must work within a hardware environment.

The real questions are:

1. With Giga 3.0 and the current hardware available using Windows XP 32-bit, what is the polyphony possibility?

2. And at what cost so that we know the financial range of practicality?

It's great to talk about Raptors, but lets not forget the 15000RPM SCSI drives available. A 73GB 15000RPM Cheetah drive costs $530 (my cost). For 146GB in either a RAID or with just a D and E drive, we're talking $1060 for two drives and at how much polyphony for the cost? Plus at 240GB for just the Pro Edition, you need bigger drives.

I think that for now, customers are advised to get systems with the existing SATA 160GB drives in a RAID and then WAIT to see what really transpires on release.

You can always change drives later once the real performance specs have been tested and published.

Peter Alexander
www.truespec.com
310-559-3779

@Nick Batzdorf said:

My understanding is that disks of all sizes cost the same to make. The difference is that the yield is lower for larger ones. (Which of course negates my first sentence, but the point remains!)

Not exactly. Higher capacities demand higher precision these times. 3.5" are 3.5" since a few years, the things that cahnged are the magnetic tracks coming together more and more. It's indeed easier and less expensive to produce with less precision. The speed increases with the packing, but also is lost on correction of read errors. Apart from that, heat problems that may arise with increasing the spinning speed could also be part of the issue.

Peter, I don't think you will get problems with Tascam. Also with 2.5 the 160 voices couldn't be reached on every system. I myself still think it's not worth the cost of the faster drives, 2 or 3 7200rpm drives should give enough polyphony, as long as there is a limit to the maximum of 2 GB RAM for precaching the files. Still I'd like to see some experiments testing the preformance when prebuffering is set lower than it is right now in 2.5 (I think 64kb per stereo sample) - but I guess therefore HD seektimes significantly lower than 9ms are needed.

PolarBear

@cm said:

one thing that doesn't work at all for streaming is to use the type of onboard raid controllers from promise (and others) for creating a raid,

Christian, does that also count for the added pci ide raid controllers by Promise? For example the Fasttrack TX2000? Or do you only mean the raid controllers built onto the motherboard?

Bests,
- M

mathis, i haven't tested the fastrack family, because this type of raid controllers is not very *intelligent* and there are lots of different types of chips and firmware used - you need to check every single piece on it's capabilities. the onboard type is unusable, at least that has been my experience.
christian

It would be helpful to know in which way you experienced these onboard controllers as unusable.
I experience some odds with my Promise and would like to know if this is similar to what you experienced.

copy a really big file from a volume on the controller to another disk to calculate the MB/s - this should be 20 MB/s or more. then try to stream a lot of voices from the same volume (1 stereo 44.1 16 bit = 176.4 KB/s) to find out how much throughput the controller allows for streaming. the first time i tried this i've been very astonished to find out it was 1 MB/s or even less. this seems to apply to all applications loading dynamically only parts of files directly from a disk instead from a cache.
christian