At 17:34 09/20/2005 -0700, Tracy R Reed wrote:
>Gus Wirth wrote:
>> the double buffering. Data transfer rate is about 25MB/s.
>> 
>> Looks like IDE drives are just slow no matter what you do. It was a fun
>> experiment though.
>
>How do you figure that 25MB/s is slow? And for this sort of end-to-end
>disk copying test I would imagine that the transfer speed has a lot to
>do with how fast the platter rotates. That doesn't have much directly to
>do with the IDE technology itself.

I call the ATA drive slow based on the results I get compared to the
advertised specifications for the drives. My drives are IBM IC35L060AVVA07
60GB 7200RPM (aka GXP60 Deathstar). The advertised sustained data transfer
rate is 47MB/s while my results show a little more than half of that.

My tests don't show the typical use pattern for hard drives and is not what
I wanted to test. There are many things going on in the system and hard
drive read/writes will be very different from what I showed.

You are correct in assuming that the rotational speed has a big influence
on data transfer rates. Faster speeds mean lower latency and faster
per-track data transfer. But the overall data transfer rates are also
highly dependant on track-to-track seek speeds, and average track seek
speeds (for some definition of "average"). ATA drives have about twice the
seek times as SCSI drives and an order of magnitude worse error rates than
SCSI drives.

I'm being a bit pendantic here to help out the lurkers. I know you (Tracy)
already know this stuff. Maybe it's time I gave my hard drive lecture again.

Gus


-- 
[email protected]
http://www.kernel-panic.org/cgi-bin/mailman/listinfo/kplug-list

Reply via email to