The boot and main storage drive in my Quicksilver is a 120GB IBM 120GXP I've had for about two years. Its an ATA100 drive but I've got it hooked to the G4's onboard ATA66 controller. I get about 22MB/sec (Xbench) from it, but obviously I could be doing better! When I looked up the G4's specs yesterday and saw it was only ATA66 I was stumped for a few mintues...why would I buy this drive and not take full advantage of it? Then I remembered! I'd bought it waaaay back when I was trying to run 10.2 on a series of idioticly upgraded pre-G3 Powermacs. That was really a pain, I can't believe I still liked X enough to get a G4! Anyway the drive had been hooked to a Sonnet Tempo Trio at the time which had to go once I decided to get a real G4.
So now I want to get this drive running at top speed again. Is anyone running anything similar to the three options below?
- ATA133 PCI card + IBM 120GXP
- ATA133 PCI card + 2 IBM 120GXPs and OS X RAID
- ATA133 HW RAID PCI Card + 2 IBM 120GXPs
All should give me some increase. The ATA133 cards have fallen to around $65. The ATA133, second drive, and software raid would run me around $150. The same option with a hardware raid card would be a bit over $200.
Barefeats.com tested setup #3 using these exact drives and got 110MB/sec - a big increase over my 22!
Can anybody post some results from 2 ATA133 drives using OS X's built in RAID software?
I don't have any reason to dump the 120GXP since its working fine. But rather than buying another should I switch to two newer, smaller, faster drives and move the 120GXP to another system or use it extra storage?
Go with a single SATA drive and add a second later?
Hope two 73GB 15K SCSI Cheetahs mysteriously arrive on my doorstep?
thanks!
-Tom
ATA66, 100, or 133 is the bus speed limit. The drive itself is doing all the speed it's capable of... even a good new ATA100 7200 RPM drive will do only roughly 40 MB/sec.
Apple's software RAID will nearly double the speed on all the drives I've tried it on except some really old ones... you might lose 10-20% of the total speed.
Ok, what follows might be a misunderstanding that's making me think this drive is capable of better:
ATA66, 100, 133 is the bus speed limit, meaning its the peak speed at which data can move from the drive. An ATA100 drive can have a peak transfer rate of 100MB/sec (peak, not sustained, average rate). When that drive is on an ATA100 controller, data can move from the drive at speeds from 0 to 100MB/sec. Resulting in an avg speed of say 40MB/s.
Stick that drive on an ATA66 controller and you limit peak data transfer to 66MB/sec even though the drive could at times deliver more. This results in a lower average transfer rate.
IBM lists the 120GXP's Avg Transfer Rate to be 23-48MB/sec.
Am I just spewing nonsense above? When I'm wrong I like to find out exactly where things are going awry. That way I don't continue building on a shoddy foundation
At least when I say things that are wrong they're based on a wrong relationship between facts, and not because I just make it all up as I go along - though that can be fun in some cases too 
Its good to hear the OS X RAID works that well! Still debating on whether to go for the hardware card or not. Its nice to know they're all made by Acard
I buy enough used stuff that warranty and support really don't matter that much on a purchase like this so I'll go for the cheapest one I can find at the time. Looks like a few more Mac items have to go out before any more come in though.
-Tom
I am not well versed in the ATA conundrum you pose, but I purchased an ATA 100 ACARD based on reading some articles stating that there was no REAL speed improvement btwn the 100 and 133 cards based on the price, when running 7200 drives 100-133.
I know this is of no REAL help, but I would suggest googling for info. From what I read, the drives sustained speeds were the "same" for both cards. This ATA 100 is in a G4 QS, but I never cared to test the transfer rate, as it was clearly faster than the onboard bus IMHO.
Athough, based on casual observations, it didn't make a difference ( I only KNEW it was faster).... Unless you are a high end user. Audio and Video....
Cheers.
VectorWorks, FormZ , and sometimes gamer.
There may be some slight chip or driver speed improvement with a newer controller, but I don't do "E.S.P." so I don't really believe it.
RAID should be obviously faster than a single drive though.
I use QuickBench from Intech's SpeedTools for drive speed testing. It gives you both sequential and sustained read and write speeds for small or large copies.
With a pair of faster drives on an ATA66, you might be knocking off some of the peaks... but I don't think you'll even get close with just one drive... or even a pair of DeskStars.
The main benefits of the ATA133 cards (other than the bus width, which would be hard to use all of with just two drives) are the dual controllers and the larger volume size limits of the ATA6 standard (larger than 128 GB in one volume).