HD Benchmarks on SCSI vs IDE

[Executioner]

Hi gang,
I have a question regarding hard drive benchmarking. The program that I used was Sisoft Sandra 2001. First my system:

Supermicro P6DGU mobo with 3 build in Adaptec SCSI controllers: Ultra2 LVD/SE, UW, and standard SCSI (50 pin).
Dual PIII-550MHz cpu's
328 megs of PC100 ram
Win2k pro with service pack 2
Seagate ST39175LW
Maxtor ultra ata 100 40 gig hard drive
Promise ata 100 controller pci card

HERE are the specs for the Seagate ST9175LW SCSI

When I ran Sisoft Sandra, it reported my SCSI drive (boot drive) with a rating of 12,271. When I performed the same test on the Maxtor IDE drive, it reported a rating of 24,740. I thought my Ultra2 drive would be close to the Maxtor. Why is there a difference? Is Ultra2 not that good?

Breakdown of test: Maxtor ATA100 40 gig drive
buffered read=69 mb/s
sequential read=35 mb/s
random read=7 mb/s
buffered write=35 mb/s
sequential write=35 mb/s
random write=9 mb/s
access time=~7ms

Breakdown of test: Seagate ST39175LW
buffered read=8 mb/s
sequential read=17 mb/s
random read=6 mb/s
buffered write=7 mb/s
sequential write=17 mb/s
random write=5 mb/s
access time=~7ms

Is Sisoft Sandra a valid test?

[PreDatoR]

Do a search on google for hdtach and umm find a patch file so it will work with NT. Its much more accurate... Ultra2 should beat a ata100 drive by a decent margin

[Jim Z]

Breakdown of test: Seagate ST39175LW
buffered read=8 mb/s
sequential read=17 mb/s
random read=6 mb/s
buffered write=7 mb/s
sequential write=17 mb/s
random write=5 mb/s
access time=~7ms

looks like it might be stuck in SE mode (single-ended), which would make it run as an UW drive. Are you sure you have 1) the proper cable and terminator, and 2) the drive isn't jumpered to force SE mode?

edit: but the buffered read rate sounds way too far off to be just that. have you tried running the test multiple times?

[nexus_7]

sandra never reported my scsi drive proformance right. I jsut gave up 2 years ago. sorry, but IDE is slower. its Killed in the access time and other stuff.

Greg

[Executioner]

Maybe it's not terminated correctly, as I never even looked. I simply verifed the ID#, installed it, and formatted the drive because it worked. I can't look right now as the pc is being used to encode a movie, but when it's done, I'll take a closer look at it. I know when it boots, Adaptec reports it as an 80 mb/s drive if that helps.

[Jim Z]

Adaptec reports it as an 80 mb/s drive if that helps.

well, then you're fine as far as cable and terminator, if you had an SE terminator it would have reported 40 mb/s. Perhaps Sandra is just brain-dead. chances are that hdtach would be a better way to go.

[DocSilly]

Argh, where should DocSCSI start to clear up this mess ???

- Sandra is a good benchmark for overall system performance but it's NOT any good for hdd benching.

- You compare a 3 year old entry level SCSI drive (it's an old 7.200rpm drive, not even a 10k) with a new IDE drive ... btw, which Maxtor model is it?

- The Barracuda is a 3.6 GB/platter design ... the Maxtor is, depending on model, most likely a 20.4 GB/platters design, maybe it's even a 33 GB/platters . Alone this makes a huge difference. It gives a huge advantage in the STR tests (sustained transfer rate).

- The SCSI has a 1MB buffer, the Maxtor most likely a 2MB buffer

- The only advantage of the SCSI drive might be it's still low seektime of 6.9 ms vs the ~ 11 ms of the Maxtor but that won't make a huge difference with drives so many generations apart.

SCSI HDDs aren't faster than IDE by default, especially newer IDE drives will easily beat older SCSI drives. SCSI vs IDE in the same generation will show an advantage for the SCSI drives (performance ... not price).
The fastest SCSI certainly stomps the fastest IDE into the ground but this comes with a huge pricetag.

The ideal setup for the performance user is a small and fast SCSI bootdrive plus an IDE (RAID?) for mass-storage. The WDxxxJB drives with 8MB cache are also nice and they come close to 10k SCSI performance in overall speed.

You should check http://www.storagereview.com if yer want to learn how the pros benchmark hdds.

[Executioner]

Thanks Doc SCSI. I was hoping you would chime in on this thread.

I use the SCSI as a boot device only. Programs and such are stored on this SCSI boot drive. All other activity is done with the Maxtor. The purpose of this system is to encode movies (i.e. copy DvD's). Since I just added this new 40 gig Maxtor (sorry, but I don't have the exact model number), and installed a Promice IDE controller cord to take advange of the Maxtor, I was a little surprised to see these numbers for this old SCSI drive. I do have a copy of HDTach, and when I'm finished encoding my movie, I'm planning to see what numbers it comes up with.

Thanks for your input.

[DocSilly]

You can rest assured that the fastest available SCSI drive will destroy the fastest available IDE drive ... at some serious cost ... the transfer rate of the Cheetah X15.3 starts above where the WD1200JB ends ...
It's the WD1200JB: 29.2 MB/sec - 48.8 MB/sec
vs Cheetah X15.3: 51.1 MB/sec - 76.4 MB/sec
The Cheetahs access times are half of the WD's ... and the idle noise level is equal.

You might be disappointed with your Barracuda numbers but SCSI drives still have an edge on IDE drives when you compare same generation drives.

[Executioner]

I ran the HD Tach program (v 2.52), and here is what I got:

Seagate SCSI:
Random access: 11.3ms
Read Speed: 16785 kps
cpu utilization: 1.4%

Maxtor IDE ATA100 40 gig:
Random access: 12.6ms
Read Speed:34941 kps
cpu utilization: 3.3%

I did not do a write speed test. The Seagate Baracuda is on FAT32, while the Maxtor is using NTFS (if that makes any difference or not)?

I do have an Quantum Atlas 10k 9.1 GB Ultra 160 drive in my closet I picked up a few months ago. It's on hold until my existing Deathstar decided to quit working. I'm planning on using it as my boot drive, and then an IDE for data.