[CLUE-Tech] hdparm

Tim Russell tim.russell at ilg.com
Thu Dec 21 08:51:01 MST 2000


Did you reboot between tests?  If not, your results came out of the buffer
cache...

Other than that, congratulations!  You survived hdparm.  It's very cool and
can yield great results, but I definitely recommend shutting down all
non-essential services, syncing your disks, and basically chanting while
incense burns before testing - it's /very/ easy for different disk
controllers to barf when you start messing with it.

If I were you I would still do a stress test - compile the kernel, etc to
make sure that your settings result in a reliable system.

Tim

P.S. Hello everyone, I just moved here from Omaha a couple of months ago.
I've been using Linux since 1994 and ran an ISP on it for three years.  Hope
to make a meeting in January if you have one!

> -----Original Message-----
> From: Roger Frank [mailto:rfrank at rfrank.net]
> Sent: Wednesday, December 20, 2000 21:23
> To: clue-tech at clue.denver.co.us
> Subject: [CLUE-Tech] hdparm
> 
> 
> Adjusting disk accesses with hdparm yield amazing benefits.
> Initially, I was getting about 3 megabytes/sec reported
> from a `hdparm -t /dev/hda` test.  I then did a
> `hdparm -d 1 -m 1 -c 1 -X66 -k 1 /dev/hda` and I'm
> getting 24 megabytes per second with the same test.  That's
> 8 times as fast.  How can this big a change not break 
> something?  Are there any tests that say  "yes it can
> go that fast and yes it is reliable at that speed."?  That's
> actually faster than I'm getting on my other system
> (with same CPU/motherboard) with Ultra-wide SCSI.
> 
> -- 
> Roger Frank
> Ponderosa High School, Colorado
> _______________________________________________
> CLUE-Tech mailing list
> CLUE-Tech at clue.denver.co.us
> http://clue.denver.co.us/mailman/listinfo/clue-tech
> 



More information about the clue-tech mailing list