First post, by auron
decided to test some of my SCSI HDDs, this is how a 1999 IBM DNES and a 2000 quantum atlas V perform in speedsys (test setup is listed there), connected to a 2940UW:
atlas V:
IBM DNES:
very different results here. i'm really not sure what to make of the extremely poor linear verify speed on the atlas drive, which also caused the benchmark to take much longer to complete than on the IBM drive. what's rather strange though is that in a past test with a pentium 60 on the same 2940UW and this IBM drive, i got ~31800 kb/s instead of the almost 29000 kb/s here, which isn't huge but repeatable. given the much newer platform, and newer/better PCI implementation with less latency to boot, one would rather expect the opposite...
i did also try replacing the 2940UW with a 29160 to see what these drives can do on a newer interface. the IBM drive is stated to run at 80mb/s by the SCSI BIOS, and 160mb/s for the quantum drive.
atlas V:
IBM DNES:
not really huge surprises, the buffered read speed numbers do see a decent improvement, but 50mb/s is still merely half of practical PCI bandwidth. maybe 16-bit access in DOS is a bottleneck here. it is noteworthy how linear read speed is helped a little bit by the 29160 on the quantum drive, but is unaffected on the IBM drive.
i should note that both drives passed the verification in adaptec BIOS, but still of course i can't be fully sure on the state of used hard drives. i've had no luck in getting any sort of SMART readouts on SCSI to work with any tool i've tried, even went to specifically update the 2940UW BIOS for this, which claims to add support, but no difference. in any event, as most here probably know, the IBM drives from the time are extremely loud and basically unusable beyond such quick benchmarking, this one is no difference. the spinning sound on the quantum drive is much more bearable in comparison.
if anyone happens to know where these drives fall into the product stacks of the time i'd be curious as it's hard to find much info on them. there's apparently a 10k RPM atlas II drive also from 2000, which seems like some confusing naming, but perhaps that one would be faster than the atlas V here.