Which is faster? ide raid0 or single sata?

[StoneNewt]

You need to give the porn sites a break and hit some computer sites instead.

Actually I've never found Raptors overly impressive, my first 36GB got aced by a 5 year old design 36GB 10K SCSI, my WD740DGs can barely out perform the same drive & all three of those drives get beaten to a pulp by my Savvios.

Before you go oh but SCSI costs so much more, well after you've laid out for your SCSI card (100£) my Maxtor (Quantum designed) 36GB cost 3£ LESS than my 36GB Rapt, both of which were a few £ more expensive than the 74GB Rapt. The Savvio how ever was 3x the cost so no surprise it walks away with the prizes.

[TooNice]

Ah, is that so? I must have neglected to check them when I did my spree of drive checking after I found out it wasnt just maxtor doing the 1 year thing. This is very good. I know the only company I'll be bothering with for hard disks from now on then.

Last time I checked, most drives with 8MB (or 16) buffer come with a 3 years warranty. If its changed, then it must be -really- recent.

[specofdust]

Well, Maxtor had switched to 1 year warrenties, then I was talking about it to a guy I LAN with, and he said all the major companies have, which was partly why he chose a raptor(since they still come with 5 year warrenties). I checked out tekheads drive descriptions and I was sure most came with only 1 year warrenties. I'd be happy to be proven wrong.

[TooNice]

I am pretty sure Maxline series is 5 years (that is one of the reason I seriously considered it).
Overclockers.co.uk states 3 years for DiamondMax, 5 years for Maxline. Looking at the other drives, its pretty much as I described. 3 years for drives with 8/16 MB cache.. sometime 5 years (Seagate, Maxline, and I believe the WD-RAID series).

I have heard about manufacturers considering to switch a while back, but it didn't last long.

[Butcher]

Actually I've never found Raptors overly impressive, my first 36GB got aced by a 5 year old design 36GB 10K SCSI, my WD740DGs can barely out perform the same drive & all three of those drives get beaten to a pulp by my Savvios.

Before you go oh but SCSI costs so much more, well after you've laid out for your SCSI card (100£) my Maxtor (Quantum designed) 36GB cost 3£ LESS than my 36GB Rapt, both of which were a few £ more expensive than the 74GB Rapt. The Savvio how ever was 3x the cost so no surprise it walks away with the prizes.

I agree. Prefer my Atlas 15k II, it's stupidly fast, and not that noisy since it's a FDB drive. 5 year warranty too, being scsi.

[TooNice]

@StoneNewt: Outperform in what? Raptor vs Savvios
Pretty close call if you ask me. I am actually inclined to give it to the Raptor, although even in the worst case scenario, the Raptor isn't exactly getting stomped.
I still consider the Raptor a premium desktop HD. It does much better than other IDE/SATA drives for server applications, but it is not on par against SCSI drive in that domain (although it won't lose as badly as other drives). For desktop applications, it compares quite favorably against most 10k RPM SCSI drive. The Atlas 10k V still has it beat, but it ain't a clean sweep.

What I am getting at isn't that the Raptor > *. But it provides a welcome performance (and price) bridge between SCSI and non SCSI drive.. Plus, SCSI cables are pretty old school (flat but wide like IDE) last time I checked, I didn't like having to deal with them.

Now if you have no doubt, and need to build the fastest PC ever, then by all mean pick up the Atlas 15k II [5dB louder than the Raptor btw] I've not won the lottery jackpot yet

[StoneNewt]

@StoneNewt: Outperform in what? Raptor vs Savvios
Pretty close call if you ask me. I am actually inclined to give it to the Raptor, although even in the worst case scenario, the Raptor isn't exactly getting stomped.

Every Raptor I've benchmarked looked good on paper, however when I start to use them things seem less responsive than a SCSI based system. When I look at it in detail and do some stats of every day work (say time opening OOo/PhotoShop, maybe to install an application from an image on the target drive or I/O bound data reduction) the Raptors always seem just not to perform at all, this is cross various SATA controllers etc. All the Savvio requires to seemly beat the Raptors is a ultra80 SCSI controller or better.

[Butcher]

Now if you have no doubt, and need to build the fastest PC ever, then by all mean pick up the Atlas 15k II [5dB louder than the Raptor btw] I've not won the lottery jackpot yet

They aren't that expensive (though I already had a scsi controller).
And I can forgive 5dB for performance like this: http://www.storagereview.com/php/ben...1=267&devCnt=2

[TooNice]

Every Raptor I've benchmarked looked good on paper, however when I start to use them things seem less responsive than a SCSI based system. When I look at it in detail and do some stats of every day work (say time opening OOo/PhotoShop, maybe to install an application from an image on the target drive or I/O bound data reduction) the Raptors always seem just not to perform at all, this is cross various SATA controllers etc. All the Savvio requires to seemly beat the Raptors is a ultra80 SCSI controller or better.

While I agree that benchmarks (especially those that just throw an HDTach/Sandra score) is not the mean of all. I must question if some of those perceived responsiveness change is "real". How many people have actually tried to use the same system, configured in the same way (where it can be, obviously you need a SCSI controller etc. on the SCSI system) and time the time it takes to open a 100MB photoshop image or whatever you fancy. Mind ya, the Savvio does have a lower access time, so I am not ruling out the possibility that you could be right.

Still, I consider the StorageReview methodology to be relatively sound (though the application they run are getting rather old - but they now have a new Testbed - I don't think it is implemented in the performance comparison database yet though). I -think- that their office drive benchmarks do take application launch into account (check the 2nd link below).

It is interesting to note that some of the fastest drive take a hit, and the slower drive get a boost.. Thats partly why I am more interested in reliability than trying to get the very best performance (especially for large IDE/SATA drives):

- http://forums.storagereview.net/inde...howtopic=20478
- http://forums.storagereview.net/inde...howtopic=20450

[SHISH]

Scusi wins on seek times as well as transfer but like all disk systems, it depends how you set them up and what hardware you`re using and for what purpose you want to use it.
A well set up system will consider several things including how many drives you have in the raid, what the block size is and whether you want sequential speed or large file transfer or many small files.
Lots of things to consider for a proper setup.
Most of the Pata/Sata drive manufacturers are now back on 3 or 5 years warranty and some do better if you specify for a server raid system or give you better MTBF warranty.
Personally, I use raid 5 on a hardware card with 4 drives per set, 8 drives per system on separate channels for a mix of reliability and speed plus losing a drive (not happened in a few years) means no data loss and a reasonably fast auto rebuild.
Only limit is dependent on the available bus speed, stripe size and block size which all need setting up according to what you want of the system. Taking winblows buffering into account can make raw speed look better than it is and a lot of people don`t even realise it`s there.

[Nox]

the above raptor links, are pointing to the slower raptor, the 74GB with TCQ enabled. They have tested raptors without TCQ, and they are a bit faster (not much). Just to confuse matters, WD also ship an 80GB raptor, as well as the 74GB. I'm currently torn between buying a single large SCSI drive, or a raptor array. hmmm.

Have to admit, I thought my two 15k scsi drives felt slower than my two raptors, but apparently that was because they were going through an LSI megaraid controller. The certainly benched slower raided, and were only barely slower non-striped.

Nox