WARNING: 386/ix 2.0 floating point problems

[Chris Wiener]

There exists a problem in Interactive's 386/ix Release 2.0 which will cause
machines equipped with the latest mask revision of Intel's 386 (a.k.a.
80386-DX) to hang whenever floating point operations are attempted.  I have
spent over 2 months working with ISC's tech support department in N.H. and
have not heard anything concrete regarding resolution of the problem.

My system consists of a Dyna 386/20 motherboard (C&T chipset), 4MB RAM, and
no numeric coprocessor.  It's performance under 1.0.6 has been flawless.
This problem occurs on 2 different machines here, and according to ISC has
been reported to them as occuring on at least 8 other systems of different
types.

The implication here is staggering!  Intel currently ships only the "D"
mask processor.  This means that if you are buying a NEW '386 box without
a '387, DON'T BUY 386/ix 2.0.

Unless Interactive wants to end up like Microport, they had better get off
their butts and fix this.  It has been 2 1/2 months since I reported the
problem, and tech support in N.H. seems to have no idea of what is being done
to fix it.

It took them 2 months to get a 80386-DX chip!  According to Tech Support,
this problem was a "priority one" bug.  Is this how ISC reacts to FATAL bugs
in their products?

####

DO NOT BUY INTERACTIVE 386/IX 2.0 IF YOUR MACHINE HAS A 80386-DX PROCESSOR
AND NO MATH COPROCESSOR!!  THIS PROBLEM IS NOT UNIQUE TO A PARTICULAR
MACHINE.

####

Since Tech Support has no idea of the status, maybe someone technical (hi
Darryl!) could provide the net with a progress report.

The more I deal with them, the more they start looking (and acting) like
Microport.

-- 
Christopher Wiener N2CR                        CR Labs, Cliffside Park, NJ
DOMAIN: cwiener at CRLABS.COM                     UCCP: ..!killer!crlabs!cwiener
WORK: 1-800-631-2524 x254 (9am-5pm EDT)