[comp.sys.ibm.pc] Weitek 3167 "Abacus" Numeric Coprocessors

b2wild@ucscb.UCSC.EDU (-b2wild-) (05/02/90)

I just bought a 3167 Weitek math co-processor for my '386 clone.
I stuck the thing on the board, but it doesn't seem to be doing
anything (the board cmos setup can't find it).  No jumpers are
listed for it on the board (just a '387 enable, which works fine
with the '387 that was replaced, but locks up the machine with the 3167).

Is there something I should be doing to activate it?  Right 
now, it seems to be just along for the ride.  Has anyone else
had trouble getting it to work?

			- b2wild -

medici@elbereth.rutgers.edu (Mark Medici) (05/03/90)

In article <2988@darkstar.ucsc.edu> b2wild@ucscb.UCSC.EDU (-b2wild-) writes:
>
>I just bought a 3167 Weitek math co-processor for my '386 clone.
>I stuck the thing on the board, but it doesn't seem to be doing
>anything (the board cmos setup can't find it).  No jumpers are
>listed for it on the board (just a '387 enable, which works fine
>with the '387 that was replaced, but locks up the machine with the 3167).
>
>Is there something I should be doing to activate it?  Right 
>now, it seems to be just along for the ride.  Has anyone else
>had trouble getting it to work?

The 3167 is not a substitute or replacement for the '387 chip - it is
an alternative.  Software that requires a '387 will not use the Weitek
3167 instead, and your system will not recognize it as a '387.  It is
for this reason that many (most?) 3167's are shipped on a daughter-
board that also permits simultaneous installation of a '387 chip.  In
order to get any use of the 3167, you must run software specifically
written to address it.

At least, this is what I recall from my research on the subject.
----------------------------------------------------------------------------
Mark Medici/SysProg3 * Rutgers University/CCIS * medici@elbereth.rutgers.edu
----------------------------------------------------------------------------