performance of nvidia onboard raid1 vs. gmirror

[Ken Gunderson]

On Wed, 03 May 2006 05:00:12 -0400
Mike Jakubik <mikej at rogers.com> wrote:

> On Wed, 2006-05-03 at 02:14 -0600, Ken Gunderson wrote:
> > Is anyone aware of any benchmarks comparing the nvidia raid1 to
> > gmirror?  Or have any additional insights?  I don't have a machine free
> > I can test on at the moment but might in next week or two if anyone else
> > is interested.
> 
> I am sure many people are interested in such a benchmark, but no one has
> done it yet. If you have the resources to do so, please go ahead and
> continue. Many of us would be interested in your results.

I've stayed away from onboard stuff since it's had a reputation of
being wonky in the past.  The comments from the dmesg just say
"impressive".  Coincidentally enough I do have a Tyan 2865 based system
configured w/gmirror on a pair of WD 74GB Raptors in a 6.0-RELEASE-p7
box sporting an Opteron 165.   So the mainboard and HD's are the same as
the dmesg poster.  

fwiw- a quick default bonnie.  In the interest of readability I've
omitted the first two lines of headings:

Machine    MB K/sec %CPU K/sec %CPU K/sec %CPU K/sec %CPU K/sec %CPU  /
sec %CPU 

selway    100 67385 56.8 50437 11.4 84770 18.6 94018 100.0
1475079 100.0 77505.9 156

Which is better than I get from dual Opteron 200 systems w/10K SCSI
drives and LSI 320-1/2x controllers (until you spawn a few instances
simultaneously..;-)

-- 
Best regards,

Ken Gunderson

Q: Because it reverses the logical flow of conversation.
A: Why is putting a reply at the top of the message frowned upon?