Click for iPhone Accessories!
Click for iPhone Accessories!


Keep this site growing - Please visit my Sponsors

Accelerate Your Mac!  - the source for performance news and reviews
The Source for Mac Performance News and Reviews
Don't forget to check out all the other site features!

Dual IBM 75GXP Drive IDE RAID Performance Tests:
By Barney Buoy
Published: 6/26/2000


[Preface: For those that have not read the past IDE RAID articles here in the last year, there seems to be some misunderstanding at least at some other web sites about IDE RAID, PCI IDE cards and Mac RAID software. IDE RAID can't run on the onboard IDE, it is only possible with PCI IDE cards since they appear as SCSI devices (and the drive drives attached to them) to the MacOS. Acard has a hardware IDE RAID PCI card (now also sold by Sonnet as their Tempo RAID card) that makes two drives appear as one to the MacOS - no Softraid drivers needed. If you don't own the Acard/Sonnet Hardware RAID card (the hardware RAID card has switches on it), then you have to use Softraid or FWB drivers for RAID. Although Softraid, FWB RAID, etc. note they support only SCSI, they will install on IDE drives attached to PCI IDE cards as noted here many times in the last year. See the Related Links at the end of this article for more info, including past performance tests with up to 2 PCI IDE cards and 4 drives. There are also installation guides for drives and Softraid in previous articles here.

Despite some confusion over VST's latest ATA/66 controller and Softraid bundle, the VST card is not a hardware RAID card (as Acard/Sonnet sells for about twice the price of the normal dual channel ATA/66 card). VST ships Softraid software for configuring drives as a RAID stripe, but it's not the same as a hardware RAID card. To clear up any confusion, all Mac PCI Cards have been dual channel (Promax Turbomax ATA/66, Acard ATA/66, Sonnet Tempo (same base card) - even the Promax Turbomax ATA/33 cards are dual channel). Due to the limitations of the interface design, IDE RAID works best when each drive is the master on a separate channel. IDE RAID in older Macs (pre-G3 series) offers little performance from what I have seen over a single drive, due to the older Macs's PCI/Memory controller I suspect (and lower overall system performance/memory/bus speeds).

Now on to Barney's report on dual drive IDE RAID with two IBM 75GXP 45GB drives.(These drives format to just over 42GB each.) Benchmark numbers are one thing - but note his real world tests in Photoshop were fairly disappointing. I expected better results. However as firmware is updated and in faster systems, results may improve. Remember the rates in pure benchmarks like ATTO Tools are usually many times higher than you'll see in real world tests. (Compare large file copy times in the Finder with the sustained rates reported by ATTO's test - usually the real world through-the-file-system tests will show transfer rates (MB/sec) many times lower than the pure benchmarks. (I.E. copy a large file, time it with a stopwatch, then divide file size in MB by the seconds it took to get the effective real-world transfer rates)-Mike]

System/Hardware Used:

Base System: G4/400MHz Yikes (PCI), 40GB Maxtor DiamondMax Plus 40Gb 7200RPM ATA66 hard drive, 20Gb Seagate Barracuda 7200RPM ATA66, 384Mb RAM, VM Off

RAID: 2 x IBM Deskstar 75GXP 45GB DTLA 307045 striped as a single volume using SoftRAID 2.2.1. ACard ATA/66 PCI IDE Controller with latest Firmware and ATA66 Ribbon cables

Installation Process:

I had a bit of trouble installing the drives. Trying to be a wise guy and as the RAID was only for testing purposes, I used a second U shaped cradle, now standard in G3 Rev 2 and G4's, but it would not fit in the far left hand side drive bay. With a bit of modification to the cradle, I'm certain it could most likely be fitted. Instead, I simple placed the cradle sideways inside the G4 case, making certain I could close the case without obstruction. [Note - for installation of two drives in the standard case for RAID - see my previous IDE RAID Installation Guide-Mike]

The drives run slightly warm. Noise is not particularly noticeable.


Benchmarks:

ATTO Tools Benchmark Test Results with Softraid Drivers:
IBM Dual 75GXP Results

[For comparison - here's a single 30GB IBM 75GXP Drive in a G4/400 AGP system onboard IDE-Mike]

*Single* IBM 75GXP 30GB Drive Results:
Single IBM 75GXP 30GB Drive Performance


Intech's QuickBench Results:
Intech Quickbench Results


MacBench Disk/Pub Disk Results:
Mac Bench Disk and Pub Disk results

[Note: these graphs will be replaced soon with Barney's tests comparing a single IBM drive to the dual drive RAID in the same system. Until then I wanted to show some sort of single drive comparison.-Mike]

MacBench Disk Inspection Tests:
Macbench Disk Inspection Tests


FWB Benchtest results w/FWB RAID Drivers:
FWB Benchtest results


Real World Tests:

Duplicating a 500Mb VPC image file

40Gb Maxtor DiamondMax Plus 40 - 0:40
86Gb IBM RAID - 0:28


Photoshop 5.5 110Mb File 99 Layers:

For these tests, as well as opening the file from the Maxtor, the file was also copied to the RAID drive. Firstly the Maxtor was used as the scratch disk, then the RAID was set as the scratch disc. Results are from opening the same file from each drive with alternating scratch disks to see if there was any appreciable difference


Opening File

File on Maxtor - Maxtor Scratch/RAID Scratch
File on RAID - Maxtor Scratch / RAID Scratch

40Gb Maxtor DiamondMax Plus 40 - 1:07 / 0:58
86Gb IBM RAID - 0:58 / 0:56


Flattening:

File on Maxtor - Maxtor Scratch/RAID Scratch
File on RAID - Maxtor Scratch / RAID Scratch

40Gb Maxtor DiamondMax Plus 40 - 0:24 / 0:18
86Gb IBM RAID - 0:24 / 0:18


Rotating 90°Counterclockwise:

File on Maxtor - Maxtor Scratch/RAID Scratch
File on RAID - Maxtor Scratch / RAID Scratch

40Gb Maxtor DiamondMax Plus 40 - 0:02 / 0:02
86Gb IBM RAID - 0:02 / 0:015


Gaussian Blur 50:

File on Maxtor - Maxtor Scratch/RAID Scratch
File on RAID - Maxtor Scratch / RAID Scratch

40Gb Maxtor DiamondMax Plus 40 - 0:10 / 0:09
86Gb IBM RAID - 0:09 / 0:09


Summary Comments:

Although file duplication, copying and file opening was noticeably faster, once again - based on all test done by Mike, myself and other readers - any gain using an IDE RAID setup once expense is factored in, is highly questionable.
-Barney Buoy
Macmechanics

[As these tests show, you're not getting anywhere near twice the real world performance (or even near 50% more) from 2 drive IDE RAID vs. a single fast drive with the controller cards we've had to date. And remember that if the single drive tests had used the faster IBM 75GXP drive instead of the Maxtor, the results would have been even closer. As you can see from articles below, the 75GXP IBM drives are faster than the Maxtor Plus40 series. Sustained rates are significantly faster (as much as 20% faster). Although marketing PR may show benchmarks that look impressive, real world performance with IDE RAID may not be significantly faster than using one of the faster modern drives with the latest Mac's onboard ATA/66 IDE controllers.

One benefit often cited is that RAID allows you have to have one huge volume (much larger than a single drive). With IBM's 75GXP drives now available in sizes of 60GB and 75GB, they may be more attractive alternatives to dual drive IDE RAID using smaller drives. If your RAID software supports Mirroring, that's another benefit should you need redundancy over performance. For owners of the current Mac models with onboard ATA/66 IDE controllers however, consider if the cost of a 2nd drive, PCI IDE card/software (and losing a PCI slot) would be worth the overall gain in real world applications. A larger single IBM 75GXP drive and more RAM might be a better value. -Mike]





Related Links:


Back to XLR8YOURMAC.COM

Your Source for the best in CPU/SCSI/VIDEO card reviews, daily news, and more! 


Copyright © Mike, 2000.

No part of this site's original content is to be reproduced in any form without permission.
All brand or product names mentioned here are properties of their respective companies.

Disclaimer: Users must read and are bound by the Site Terms & Conditions of Use.