[comp.arch] 80386 32-Bit Multiply Problems?

trimble@leadsv.UUCP (04/12/87)

The San Jose Mercury News today reported on "errors in the intel
32-bit chip when doingg a multiply of  32-bit value with another
32-bit value".

The article inferred that the greatest impact would be on the Compac
company who is shipping the largest number of 80386-based equipment
currently. It was inferred that intel would replace the current part
over the next n monthss (years).

A few questions:

	1) What IS the result of this operation? can someone post
	a result (pure intellectual curiosity).

	2) Has Intel posted this to the field with a position
	statement on part replacement?

This is meant to have no bearing on Intel as a parts producer and
supplier.  Merely my (personal) interest.

I'd type in the whole article but I'm not sure if it's legal.

Gary M. Trimble - Lockheed Advanced Marine Systems
UUCP: {(ucbvax!dual!sun) (ihnp4!qubix)}!sunncal!leadsv!trimble
      {{allegra ihnp4 dual}!fortune!decvax!decwrl}!amdcad!cae780!leadsv!trimble


-- 
Gary M. Trimble - Lockheed Advanced Marine Systems
UUCP: {(ucbvax!dual!sun) (ihnp4!qubix)}!sunncal!leadsv!trimble
      {{allegra ihnp4 dual}!fortune!decvax!decwrl}!amdcad!cae780!leadsv!trimble

greg@utcsri.UUCP (04/13/87)

In article <1447@leadsv.UUCP> trimble@leadsv.UUCP (Gary Trimble) writes:
>The San Jose Mercury News today reported on "errors in the intel
>32-bit chip when doingg a multiply of  32-bit value with another
>32-bit value".
>
>The article inferred that the greatest impact would be on the Compaq
>company who is shipping the largest number of 80386-based equipment
>currently.
[..]
>	1) What IS the result of this operation? can someone post
>	a result (pure intellectual curiosity).
[..]
	3) can it be easily worked-around in a compiler  0.25 * :-)

[ I can see it now 'the startup routine does a multiply and sees if
the result is right. If so, it arranges for the multiply emulation
trap to replace the INT ops with real multiply ops, and restart them..]
-- 
----------------------------------------------------------------------
Greg Smith     University of Toronto      UUCP: ..utzoo!utcsri!greg
Have vAX, will hack...