RE: DX4-74 --> 5x86-100 Conversion Question

[Chris DiBona <chrisd@loc201.tandem.com>]

I appreciate all the input from the group regarding the 5x86 upgrade question. I was wondering if David Ross could post his impressions on the speed differences with the new chip on his desktop machine. 

  Chris

----------
From: 	David Ross[SMTP:ross@math.hawaii.edu]
Sent: 	Wednesday, December 04, 1996 2:06 AM
To: 	David Fox
Cc: 	thinkpad@cs.utk.edu
Subject: 	Re: DX4-74 --> 5x86-100 Conversion Question

With respect to claimed speed improvements if the 701 chip is replaced
with a 5x86, I wrote:

> > If PEP or anyone else is seriously
> > claiming that the new chip will double the speed, they should be
> > willing to back it up with some sort of evidence, or maybe a
> > money-back guarantee:-)

David Fox replied:

> Wouldn't that be AMD's responsibility?  I'm interested in hearing
> opinions of how big the improvement is likely to be - you sound pretty
> skeptical.

Neither AMD nor Cyrix (I don't know which chip PEP uses) make claims
about *systemic* speed improvements; this is very difficult, since
system speed is also a function of bus, cache, and BIOS (not to mention
video and drive - both relatively fast on the 701, by the way).  This is
a common topic of discussion on several Usenet groups.  Despite my
skepticism, I plan to replace the DX2-66 on my home desktop machine with
one of these chips - it is clear that the bottleneck on *that* machine
is the CPU, and anyway for $40-$50 it is worth a shot.  However, on the
701 the bus and lack of cache seem to be real problems: for example,
on most systemic benchmarks it runs much slower than my home machine
(though the disk and video are comparable).  I'd be really surprised to
see more than a marginal improvement in speed from the upgrade,
certainly not enough to warrant the cost (much more than the $50 the
chip itself costs!)

- David R.