i have been using a scsi atlas 160 as my primary audio drive with terrific
results. no problems. but now i am bulding a few little giga monster machines to be used with the vsl library. i am considering IDE since obviously it is cheaper but i have been out of the IDE loop for a while . what drives/ controller combos have been working for giga users? which ones should i avoid. what about thse ide raids are they as good as they sound?? any tips are very much welcome
thanks in advanced
Mike, I used to be SCSI only, but today\'s IDE drives are faster than the SCSI drives of just a few years ago. When I pulled SCSI out of my original GS system, my performance went up, and this system has never had any problems, either when it was SCSI based or since it became IDE based.
I believe any 7200 RPM ATA 100 drive with a 2 meg buffer (pretty much any current drive) will be fine. The best performance reviews I\'ve seen are here:
I only run drives as masters; IDE cannot use 2 drives (master and slave) on a cable simultaneously. Also if the interfaces of the drives are different, most controllers will run both drives in the lowest common denominator mode, so your ATA33 CD slave drive could kill the performance of your ATA100 master system drive. Old drives couldn\'t saturate ATA33 controllers, so you might not notice, but modern drives can. I use Promise Ultra100TX2 controllers when I need more IDE masters; these work well.
I\'d take individual drive recommendations with a grain of salt, no individual will have statistically meaningful data. That said... I have used 3 IBM GXP60 drives for a couple of years without trouble. These have a bad rap (mine are backed up!) but they are seek monsters and fairly quiet.
I\'ve used (and failed) a Seagate Barracude iv, these are supposedly reliable, said to be slow but I thought the performance was very good, and they are silent!
I\'m using a few recent WD drives without problems, these are the fastest, and moderately quiet.
I think all modern drives are plenty fast for GS. I don\'t believe IDE drives are nearly as well made as SCSI, so it\'s more important to have your backup situation in order. I don\'t use RAID, but I would never use an IDE striped array where ANY drive failure takes out all the data.
_if_ the controller on your motherboard/expansioncard and the interface of the harddisk/cdrom is well-designed and you have installed the appropriate drivers, it _should_ have no effect on the fast units.
btw: the possible throughput of the IDE-bus is much higher than the amount of data a harddrive can provide to the bus, so you don\'t have to worry about performance using 2 devices on one bus.
e.g. even ATA33 using DMA gives you 33MB/s and you will hardly find an IDE-harddisk giving you more than 15 MB/s sustained datarate
This command read a gig in 22 seconds! (My apologies to non-nerds for this arcane babble...)
Granted this is a raw sector read and you\'d expect lower performance from a file system read, but ultimately that will be capped by the transfer rate. In the real GS / win98 world, I have compared identical drives on ATA33 onboard vs ATA100 controllers, and the performance was substantially better on the ATA100 controller, simply because the ATA33 bus saturates and limits the transfer rate. I wouldn\'t expect to see any difference on a drive more than a year or two old though, they weren\'t that fast then.
In response to Bill:
> Has anyone done any actual testing on this?
I can\'t test it (no slave drives 8^) but it\'s got to be easy enough for someone here. There\'s got to be many folks here with 2 identical drives + a CD. If both drives are similarly full and defragged, you should be able to test the performance just copying a large file within that drive. If the time isn\'t close, you might have a problem.
possibly the drive you mentioned to be faster on a ATA100 than on a ATA33 has an ATA66-interface?
having the WD you mentioned installed, my calculation would be analog using an ATA100 controller
because there are only few IDE-drives here, a quick check with *old* 18GB-barracudas (ATA3 - ST320414A) on a ATA33-controller - one master on prim.IDE another master on sec.IDE with an *old* asus-DVD-drive (PIO4) as slave, both drives uDMA2, not defragmented but mirrored, 50% full, NTFS (real-life-conditions, so to speak):
random access 4,4 MB/s, sequential: 24,3 MB/s - same result if i\'m copying files from dvd to an attached SCSI-disk at the same time or not.
a second config i could test are two WD1800 and a brandnew hp-dvd-writer on an ATA133-controller, but i\'m pretty sure, the result would be similar (with higher rates of course) - i will post it
thanks for the insight and the link. i was surprised to see the results sata baracuda.
what is still don\'t understand is that on the tascam site , they advise using drives that have a access time of 9ms or less. yet many of the ata users are getting satisfacory results with drives that barely have 9 ms in some cases more.. why is this?? are there other parameters that tascam is\'nt reporting ? ie average seek time , mbps??
Mike, I will guess that \'average\' seek times aren\'t the whole story. 160 voices requires 14 megabytes per second sustained transfer (non contiguous, wowee!). That would have been tough 4 years ago, now all the drives smoke that figure. Also, drive densities have gone way up, so it\'s concievable that seek distances within a gig file have gone down considerably. I believe that average seek figures represent a seek halfway across the platter, that covers a lot more storage now.
Christian, my test of ATA33 vs ATA100 interfaces were on 2 empty IBM gxp60 drives which have ata100 interfaces. Storage review says these drives transfer about 40MBps at the beginning of the disk.
The rumors are true… some of them, at least .
StorageReview.com readers have been speculating for the better part of three years on when the industry would ratchet up the spindle speed of ATA hard drives. When would it happen? Which company would start the trend?
Today Western Digital announces its Raptor WD360GD, the first (Serial) ATA hard disk to feature 10,000 RPM operation. The Raptor features a single 3” platter that stores 36 gigabytes. Folks may recall that raptors were carnivorous dinosaurs that, while not as large as some other predators, nonetheless were swift, agile, and hunted in packs. WD envisions several of these drives in conjunction with an SATA RAID controller as a low-cost solution that can service the low- to mid-range enterprise storage sector.
That’s right, enterprise-class. Hence, the WD360GD features a very SCSI-like 5.2 millisecond seek time along with an 8-megabyte buffer. The firm also claims a 1.2 million hour MTBF spec and backs the drive with a 5-year warranty.
Price? Today’s current 36-gigabyte SCSI units run about $210-$220 from various resellers. WD aims to deliver the Raptor at around $160.
With no SCSI business to preserve, Western Digital emerges as the manufacturer to pull the trigger on higher-speed ATA units. The Raptor’s enterprise orientation, however, forced WD to wait for Serial ATA and its spec-level hot-swap ability.
When will we see these drives? In a conference call with SR, WD took great pains to emphasize that these drives are ready to go and that the official announcement was timed to minimize the window between the announcement itself and general availability. Look for a 1-2 month time frame. Don’t expect to find these drives sitting on the retail shelf next to the Caviar JB’s, however. The Raptor’s enterprise/server orientation means you’ll have to purchase it from the same specialty retailers that sell SCSI disks.
We’re aiming, of course, to get a sample or two as soon as possible. We’ll keep you updated on any further news and progress.