[plug] HDD performance

[John Summerfield]

> On Thu, 18 Nov 1999, John Summerfield wrote:
> 
> > [root at possum summer]# /sbin/hdparm -ti /dev/hda
> 
> >  RawCHS=17662/16/63, 
> 
> So thats 0.5MB/cyl at 7200rpm = 60MB/sec theoretical max at the heads,
> right? Can the drive read all heads simultaneously?
> 
> >  Timing buffered disk reads:  32 MB in  1.91 seconds =16.75 MB/sec
> 
> Yowser! I get < 4 MB/sec with my ordinary drive. Are these figures
> very meaningful?

They are. The drive is quick.

> 
> > This is a 7200 rpm drive in a PII-233, LX chipset, UDMA/33. 
> > Note the cache size;-)
> 
> What difference does 2MB vs 0.5MB make on a  32MB transfer?

Little. It kicks in for databases though.

> I'd expect it to be proportionally faster than a 5400rpm at best. Anything
> else different?

It is proportional. Newer 5400 RP drives do about 12 Mbytes/sec. Some 
respond to tweaking with hdparm, some do not. The IBM drive is indifferent 
- I can't improve its default performance.

> 
> > closely related to the rotation speed and the real track capacity (not 
> > what the BIOS  thinks or what the drive says).
> 
> So is the RawCHS quoted above correct? Do drives really have 16 heads?
I'd not beleive the RawCHS figures. My guess (I've not seen the info) is 
that the drive has four platters, possibly three.




-- 
Cheers
John Summerfield
http://os2.ami.com.au/os2/ for OS/2 support.
Configuration, networking, combined IBM ftpsites index.