[lug] RH Linux download + hdparm

[D. Stimits]

"Sexton, George" wrote:
> 
> Which brings up a really under appreciated point. Just because the interface
> is rated at a certain speed it doesn't mean the drive can do anything like
> that kind of speed on sustained reads or writes.
> 
> A good example is Ultra-160 SCSI drives. Generally, the drives cannot read
> or write more than 20-25MB second. Period. End of sentence. Look at the
> sustained throughput rate specification if you want to find out what you can
> realistically do with a drive.

Depends on density and rpm. I have U160 drives that sustain just in
excess of 30 MB/sec, both manually timed, and reported by tools (I
forget which tools, hdparm was only one tool). It also depends on
whether the reads are on the outer cylinders or inner cylinders. What
happens with U160 that is really nice is when you get one with a large
cache, e.g., 8 MB...you can more easily sustain 30+ MB/sec during random
reads this way. The measured burst rate is just over 140 MB/sec, which
counts when you have more than one drive on the line, and takes more
advantage of drive cache. There are some very large density drives (80
GB+) that are able to sustain over 40 MB/sec, some approaching 50
MB/sec. People drastically underestimate the speed value of higher
density clusters in combination with 10k or 15k rpm. Unfortunately, you
can't get 15k rpm in high density, nor on IDE. But anything 36 GB or
higher (while running at least 10k rpm) can exceed 25 MB/sec over
sustained time periods.

> 
> I just took a peek at a MAXTOR spec and see that they don't even report this
> number. I know IBM 10K Ultra-160 drives are something like 22MB/Sec. I
> seriously doubt that consumer grade drives are faster than this.

Depends on density, rpm is only a small part of this. Splitting over
more platters rather than lots of clusters on a single or just a few
platters also does not improve density.

> 
> So, the reality is that in a single drive configuration you are unlikely to
> ever see more than 25MB Sec across the bus for sustained reads/writes.

I would go so far as to suggest that average consumer drives seldom can
sustain over 10 MB/sec, above average (but still general consumer)
drives probably have trouble sustaining 20 MB/sec.

D. Stimits, stimits at idcomm.com

> 
> >> for the specs on the drive, I discovered that although the drive
> >> electronics has the ATA-100 interface, the rest of the components were
> >> from the ATA-66 drive, with a corresponding throughput.  IOW, just
> because
> 
> -----Original Message-----
> From: lug-admin at lug.boulder.co.us [mailto:lug-admin at lug.boulder.co.us]On
> Behalf Of John Karns
> Sent: 18 October, 2001 12:33 PM
> To: lug at lug.boulder.co.us
> Subject: Re: [lug] RH Linux download + hdparm
> 
> On Wed, 17 Oct 2001, Calvin Dodge said:
> 
> > On Wed, Oct 17, 2001 at 06:58:00PM +0000, Greg Horne wrote:
> > >
> > > Do the different distros on them make any differece?  Do the hard drive
> > > brands play a big role in this?  Are there any safe tweaks to speed up
> > > access times?
> 
> > The drives do make a difference - Mom's 500 Mhz K6-based system (with
> > a Tyan S1598 motherboard) has a 5 gig Micropolis and a 30 gig IBM.
> > The Micropolis maxes out at about 4 megs/second, while the IBM (a 5400
> > RPM model) does about 16-17 megs/second.
> 
> Indeed, and labels can also be misleading.  I recently bought a 40GB
> Quantum ATA-100 drive.  I wasn't so much interested in the data band width
> (bw) as much as the large capacity.  In perusing the Quantum Internet site
> for the specs on the drive, I discovered that although the drive
> electronics has the ATA-100 interface, the rest of the components were
> from the ATA-66 drive, with a corresponding throughput.  IOW, just because
> a drive has an ATA-100 interface doesn't guarantee that
> the drive fully supports the spec.
> 
> ----------------------------------------------------------------
> John Karns                                        jkarns at csd.net
> 
> _______________________________________________
> Web Page:  http://lug.boulder.co.us
> Mailing List: http://lists.lug.boulder.co.us/mailman/listinfo/lug
> 
> _______________________________________________
> Web Page:  http://lug.boulder.co.us
> Mailing List: http://lists.lug.boulder.co.us/mailman/listinfo/lug