Without first spotting this thread, I have just started a new one showing the difference I experienced going from a single fast SCSI drive to two fast SCSI drives configured as RAID 0. My impression is favourable.I suppose it might be unsound to predict SATA performance based on the results obtained on a SCSI system, but presumably the relativities should be similar, even if SATA can't quite match SCSI speeds yet.Tim Morshead

SCSI and SATA are busses. The buss speed is irrelevant unless it becomes the limiting factor. For instance, my three 7200 RPMers are transferring at a STR of ~ 300MB/s. SATA's theoretical limit is 375MB/s (3Gb / 8 = 375MB). Unless you are up against this kind of buss limit, SATA vs SCSI doesn't matter. It's drive that matters. What cha running?

SCSI and SATA are different bus technologies; in theory the biggest difference is the total available bandwidth; but you don't need a 6+ disk RAID 0 configuration to feel the benefits of SCSI. This is because the manufacturers of SCSI drives typically build them to higher specifications than SATA drives. They do so because they are targeting different markets.I dare say there is nothing intrinsic to the technology to prevent SATA and SCSI drives being built to the same specifications, but this simply has not happened yet, so far as I know. For example, I do not think you can yet buy a SATA drive spinning at 15,000 RPM, or with seek/access times as low as the best SCSI drives. This means that in practice you must use SCSI drives for the highest possible sustainable transfer rates and the fastest access times. As with most things, you could of course get it wrong - for example, by buying an older SCSI drive, or a slow one, or setting things up wrongly (eg, turning off U320 support for drives which support it). But this does not affect the general rule.The performance of some of the latest SATA drives looks tempting. For value for money, I am sure that a Raptor is a better bet. But dig a little deeper and you may find that (for example) the headline transfer rate of a SATA drive cannot be as well sustained across the full width of the spindle as with the best SCSI drives. That is certainly what I found when I examined the position in detail some years ago.By SCSI I include SAS.Tim Morshead

http://discountechnology.com/Seagate-ST330...9&category=-109Here's a nice, current production Seagate SCSI, 15K RPM, 300 gigerbiter for $575. It uses the SCSI 320 buss. This means the buss can handle up to 320MB/s. (SATA II handles 375MB/s). Now, you gotta get a SCSI PCI(e) adapter (Our consumer mobos only support IDE and SATA). These are from 200 to 600 bucks. Let's call the total cost 750 bucks. We got it all on sale. This 15K SCSI can sustain 125MB/s at a 4ms seek. A single SATA II can only sustain 100MB/s at 6ms seek. So to compare, buy two SATA II 250MB drives for $150 (total) and plug them into your existing box. Then use Intel Matrix Storage manager to Raid 0 Only 150MB of each of the 250s to create a 300GB Raid 0. Leave the rest in Raid 1 for backup. This strategy will force the Raid 0 to use only the fastest portion of the drives. Sustained xfers of 175MB/s+ are easily sustained across the entire raid. The SATA beats the pants off the SCSI in sustained transfer rates. But there's a down side. That $600 saving with the SATA Raid 0 setup gives up 2 ms of seek time. That's $300 per millisecond. It just boils down that same ol' dollars vs Sense equation. Speed costs. How fast do you want to go?

It goes without saying that the consumer technology delivers the best value for money. It would be surprising otherwise. But I don't think anyone can judge for anyone else whether it is "sensible" to spend money on a SCSI system. There are, clearly, distinct advantages for those willing & able to pay the extra cash, or who can justify the expenditure by using their PCs for other applications; and not all of them have to do with the bus.Your comparison between a SATA RAID 0 and a single SCSI drive is interesting and it does indeed help to illustrate how SATA drives give better value for money than SCSI. But your example also illustrates how the SCSI retains an advantage of 2ms in access times. Particularly for an application like FSX, which has to access lots of small disparate files as well as smaller numbers of large ones, access times are an important part of the equation. I appreciate that the difference between 4ms and 6ms is an average, but 2ms is not by any means a trivial advantage over any protracted period of use.Tim

I cannot answer the RAID question but here is a fact: just built a new system and found the 10000 Raptor to be slower than the WD 7500: WDC WD1500AHFD: 85 MB/sWDC WD7500AYYS: 94 MB/stested with Everest 4.2. the Raptor has the OS

I've attached my benchmark results of my three drives (well, the three that I use most often anyway), the key is Yellow: My Primary C: drive which is my Raptor 10k drive.Red: G: drive which is my Raid O 465GB driveGreen: D: drive which is a SATA II basic drive.Going by these results, which would be the best to install FSX on? Currently it's on the Green drive. But after re-running these Everest Disc Benchmarks I would have to assume that the Raid drive is the better option as in all read tests, it's showing the highest scores. Is this correct?In the past I've made sure to only put on Raid drives that which I can afford to lose, as in this time by reformating and going back to XP Im not sure what happened but Windows XP refused to recogonize an existing Raid drive and so the data that was on it was lost as I had to recreate the Raid in XP, unfortunatley there was actually quite a bit of data that I didn't want to lose on the Raid drive that I had stuffed on there for safe keeping (yeah, right!) when I made the switch up to Vista and then lost on the way back to XP. Oh well, I'll just have to recreate it.Anyway, so by these benchmarks, I'd assume that I should be putting my games and other applications that do a lot of read access on the RAid drive, is this correct?

FS has always used a whole bunch of small files, but is working toward a file strategy of smaller, larger files. I don't know how that's going, but that's the direction. Any HD will have to do 2 things. It will have to (1), locate the file. This is influenced by the Seek times (or "Read Access"). Then (2), move the file. This is the transfer rate (Random, Liner, Buffered Read tests).The relative importance of Transfer Rates vs Seek Times is really the crux of the discussion.For a 'small/lotsa' file structure, seek times will be Very important. Ya gotta find all those files, one at a time. For a 'large/fewer' file system, transfer rates are more important. For our current FS uses a small/lotsa file system where seek times are very important. The question then becomes "Where does strength in one one start to offset weakness in another?"For instance, the Raptor can xfer at ~ 70MB/s and can seek at ~ 5ms. My raid can transfer at 4 times that (300 MB/s), but has an access rate of ~ 6 ms. Does the 400% in advantage in access times make up for a 25% disadvantage in seek times? Well, it's hard to say no . . . plus it has the right skill-set to take advantage of the direction of FS's file system in the future. 2X the storage space, plus room to setup a redundant raid 1 on the same set of drives for the same price as a Raptor is a nice bonus, very nice. (2x250GB 7200s= $150 vs 1x150GB 10000=$150. Intel Matrix Storage manager allows the 'slice and dice' with the raid.) But tech specs are a tough way to determine real world performance. Try it out on both.

I don't really trust Everest. My Raptor is showing 8 MB/s Random Read and should have 13 MB/s according to Tom's Hardware testing. I called WD and they tell me you can't reliably test the HD with the O/S on. Also format state seems to be an issue. HD must be fully pre-formatted (not Quick Format) for the test to be accurate.

mystery solved, I got my units mixed up. it's ms not MB/s

>mystery solved, I got my units mixed up. it's ms not MB/sSo that was seek time then.??RhettAMD 3700+ (@2585 mhz), eVGA 7800GT 256 (Guru3D 93.71), ASUS A8N-E, PC Power 510 SLI, 2gb Corsair XMS 3-3-3-8 (1T), WD 150 gig 10000rpm Raptor, WD 250gig 7200rpm SATA2, Seagate 120gb 5400 rpm external HD, CoolerMaster Praetorian

The raptor should transfer at about 70 Mega Bytes (not bits) per second. New 7200 RPM drives xfer at over 100MB/s. Raptors are able to find any piece of data (seek or access time) in about 5 milliseconds (5, one thousandths of a second). New 7200 RPM-ers seek at about 6ms. These are averaged times. Any HD will be able to find data (access time) and transfer data (transfer rate) faster when it is located on the outside edge of the HD's data disk. Everything slows down as the HD uses data that is closer to the center of its data disk. That's why really helps to use a defragger that can move selected data to the outer edge of the HD's disk (like O&O defrag).The format/op system inaccuracy information is correct, however it is insignificant for real world purposes. This measurable difference will never be discernible in a real life application.