PCA Forums

After 2 IBM 75GXPs dying on me I couldn't afford to buy 2 brand new drives. I read some reviews on the new Western Digital WD1002JB and WD1000JB featuring an 8Mb buffer. WD claims the performance can compete with 15k SCSI but the reviews say it's more like a 10k SCSI. I picked one up at my local Fry's and did a quick bench comparison using Winstone. Here are some numbers -

Promise Fastrack 100 w/ 2 IBM 45Gb 75GXPs (striped):
Winstone - 9080 Mb/s
High end Winstone - 23600 Mb/s

Western Digital WD1000JB at DMA mode 5:
Winstone - 10300 Mb/s
High end Winstone - 33300 Mb/s

This is one impressive drive considering that it beats my IBMs in a RAID setup. Now if only I had one more of these WD drives to make a screaming fast RAID array.

Very nice indeed

The RAID should still beat the WD in raw transfer speed.
Do you have WinBench 99 to run some more benchmarks?

I'm just looking at storagereview.com and the low-level scores they get by using WinBench 99 (since this is also still being used in their latest HDD reviews). I'm well aware that the numbers I quote can't be compared directly since they're gathered from 2 different testbeds ... but shouldn't make a HUGE difference.

Anyways, here some numbers from SR:

- Promise FT 100 with 2x Maxtor in RAID-0
> Disktransfer:
58700 KB/sec Beginning
34667 KB/sec End
> Disk Winmark 99
6124 KB/sec Business
9424 KB/sec High End (15400 single drive)

- Western Digital WD1000BB-SE (alias WD1000JB)
> Disktransfer:
43000 KB/sec Beginning (43733 WD1000BB)
27900 KB/sec End (27900 WD1000BB)
> Disk Winmark 99
10900 KB/sec Business (8633 WD1000BB drive)
25567 KB/sec High End (22533 WD1000BB drive)
... you can easily see that the 8MB vs the 2MB of the BB drive doesn't change a thing for the raw transfer speed but certainly helps with the Winmark 99 scores

- IBM 75GXP 45GB
> Disktransfer
37200 KB/sec Beginning
19700 KB/sec End
> Disk Winmark 99
7653 KB/sec Business
19367 KB/sec High End

Some notes:
I checked the review of the Maxtor 80 and the Winmark 99 scores of the single drive (prolly different testbed than RAID testbed) and they are close to the FT100 single drive numbers.
The FT100 review does mention that the stripped High End numbers are way to low and that there might be problems with WinBench.

The IBM 75GXP single drive numbers are slightly better than the Maxtor 80 numbers but they're well below the WD numbers.

RAID-0 will only help in one area and that is raw speed in transfer rate.
RAID-0 can not speed up random access times.

This can be easily seen in the RAID reviews where the transfer rate increases in RAID-0 over the single drive on same controler while the Winbench Winmark 99 numbers remain almost the same between single drive and RAID-0.

I guess that Winstone uses a comparabe benchmarking as Winbench 99 Winmarks where they emulate a couple applications usage. This is why the WD seems to be faster than the RAID-0 ... but that doesn't include the raw speed.
The WD1000JB is also a superior drive compared with the IBM 75GXPs in your RAID. It is 2 generations ahead of the IBMs, it uses 40GB/platter vs IBMs 15GB/platter and it has the huge 8MB cache.

Conclusion (of my rambling):
Yes, the WD1000JB is faster than your RAID-0 ... looking at Winstone/Winmark numbers.
No, the WD1000JB shouldn't beat the raw transfer speed.
The faster WD will be better than the stripped IBM RAID in most of he cases unless you move a lot of huge files.

I wish I had the cash to buy TWO of those JB's and RAID 0 them !!! Now THAT would be somethin' !! Seriously tho, those are sweet drives even as a single.

I just ditched a 2 x 30GB WD 7200/ata100 RAID 0 config in favor of a single Maxtor 60GB 7200/ata133 D740X and while there is some difference in synthetic benchmarks, I don't notice any "signifcant" real world slow down.

After having a PCI RAID controller go belly up on me and thoroughly wasting my data ( thank goodness for backup !), I just decided to forego any similar future aggravation for the time being. I'm running the Maxtor as a single off the Hpt372 0/b RAID controller on my Abit KR7A-RAID mainboard and it gets better than average #'s according to Sandra. That's close enough for me.

I checked out those 100GB WD-JBs tho and did plenty of drooling whilst looking at the #s they put up. In the end tho, being a cheap SOB, the 60GB Maxtor for 100 bux less seemed like a more reasonable play for me.

yowser... that's some good info there Doc.

I would have run more benchmarks to compare the two setups but I was eager to backup from the IBMs since one was already exhibiting signs of failure. Now that I've safely recovered my data I just might run some more tests.

Besides being significantly quieter I really haven't noticed much difference with the WD1000JB. I was happy with my RAID but a little soured by my bad experience with the IBMs. Every day I felt like I was sitting on pins and needles where at any moment I would again hear the "grind grind grind grind" of a dying drive.

VidmanII, you got any numbers with the ata133 Maxtor?

Hey solstice, been awhile man.

i've been running 2 x 30.7 ata66/5400 drives for over a year on Raid 0 with no problems.

i'll be shuffling a couple partitions over to a new WD800BB (80, ata100/7200) and nuking the rest.

I'll probably run a couple before and after benchmarks to see what I can expect from this new drive.