I'll echo this with my experience with RF72s on a KFQSA.. dog-slow. 
Some 3.5" 2 gig dssi disk i have (rf74?) on a KA690 box is quite fast, 
though.. well, its running VMS so i havent benchmarked it more than 
trivially. I think both the disks and at least some of the dssi 
interfaces might be quite slow. Probably more the disks than the 
connection. Also, is this a SHAC were talking about, or an SII? 
SII is _very_ slow!

isildur 

On Fri, 22 Nov 2002, Dave McGuire wrote:

> On Friday, November 22, 2002, at 03:03 PM, Stanley Reynolds wrote:
> >>    You got better performance over the 3.3MB/sec
> >> qbus than an CPU-based
> >> controller?  That's one I'd have to see for myself,
> >> I think.
> >
> > Yes, we did benchmarks on both random and sequential
> > reads of large and small blocks. We also did
> > benchmarks on CPU performance. We also ran a mix of
> > benchmarks to measure CPU utilization, there was no
> > question that the DSSI disk were slower and showed no
> > advantage.
> 
>    Wow, that's a big surprise.  So the only advantage of DSSI in those 
> situations is its cluster interconnect capability.
> 
>        -Dave
> 
> --
> Dave McGuire              "So you're really not gonna bite on the
> St. Petersburg, FL              transvestite bait?"    -Doc Shipley
> 
>