On January 10, Steven M. Schultz wrote:
> > factor of 60 :-)  Even a 386DX/16 would likely fair well against a
> > MicroVAX II with the dhrystone benchmark. But I've a feeling that floating
> 
> 	the 386DX blows the uVax-II out of the water.

  Unless you try to do more than two or three things at a time.

> 	The "PC" (386) would be able to use more current SCSI devices (but the
> 	thought of a Cheetah drive hooked up to a 386 wobbles the mind ;-)) 
> 	whereas the uVax-II could not for the most part.

  I dunno 'bout that...I've got SCSI disks in my KA630-based
VAX...There's a lot of Qbus SCSI in the NetBSD/vax crowd.

> 	For 'raw' cpu power a 386DX/33 is equal to a (get this) a VAX-8650 
> 	(what we used to call "big iron").  It was a real suprise/awakening
> 	years ago when I benched a 386/33 and it came out faster than a 8650.
> 	Now for sure a 8650 could move a *LOT* more I/O over multiple busses, 
> 	but CPU wise (and if YOU're the only user) the 386 is faster than any 
> 	VAX up thru the 8650.

  Why is it, then, that one can do so much more work on an 8650 than
one can on a 386?  It's proof that most, if not all, benchmarks do
very little to compare real-world performance of unlike architectures.

  After re-reading your above paragraph to make sure I wasn't flaming
you *too* hard, I should give you more credit for your statement about
multiple I/O buses... :-)

  I'll say this as politely as I possibly can: The NetBSD port-vax
list probably isn't the right place to be spouting "intel boxes are
better than VAXes" stories.  Especially when some of use just woke up
and haven't had our morning Mountain Dew.


                          -Dave McGuire
                           mcguire@neurotica.com