On 7/24/2006 at 10:26 PM ard at p850ug1.demon.co.uk wrote:
Now, I've not really used the 29116, I've
only come across it in the PERQ
AGW3300 graphics processor, but it strikes me from the limited data that
I have on it, that if you use this chip you're limited to a 16 bit data
path. It's not really a bitslice chip in that you can't stick <n> of them
togehter (like you can with the 2901).
True, but there was IIRC, a 32-bit version also. AMD calls 'em
YEs, but the OP wants to make a 20 bit processor...
microprocessors, but without so much as a P-counter,
it's hard to justify
the name. Still, I wonder if a 5-package 2901 version without the carry
lookahead would be any faster than 2 trips (microprogrammed) through the
16-bit CMOS unit.
Yeah, and it would probably be even faster to simulate the thing on a
Pentium<n> for suitable values of <n>. But that's not the point, surely.
-tony