On Sat, 10 Jan 1998, Steven M. Schultz wrote:

> 	Ummm, not really.  Both are hobbled by 'crippled' bus structures - the
> 	uVax-II by the multiplexed Qbus and the 386 by the ISA bus (which
> 	BTW is faster than the Qbus ;)).

But then again, the Qbus is a much better design -- much more flexible;
scales so much better it's in a different class altogether; there are
standardized configuration rules; and you get to decide for yourself
what relative priorities you want your hardware options to have.

> 	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.

I remember, getting on 10 years back now, I guess, compiling and running
"UNIX TeX" on two different non-UNIX platforms: a 12MHz 286 running DOS,
and a VAX 8700 running VMS.  Evan when I had the VAX all to myself at
night, the PC would still beat it at TeXing documents.

Of course, when I started digging into this, and learned a bit about the
RMS file system, I found out why.  I remember increasing the "IOzone" 
benchmark results for that VAX by a factor of 100 just by adding a few
lines of C code to the benchmark source (buffering I/O to be able to use
fixed record length binary files instead of the CRLF-delimited stream
files that were the C run time default).

I sent my patches to the IOzone benchmark author, but he never replied.

-tih
-- 
Popularity is the hallmark of mediocrity.  --Niles Crane, "Frasier"