AM29116DC cpu??

From: Paul Koning <pkoning_at_equallogic.com>
Date: Mon, 28 Feb 2005 14:06:02 +0000

>>>>> "Dwight" == Dwight K Elvey <dwight.elvey_at_amd.com> writes:

 Dwight> Hi I might have a manual someplace that has it in it. It was
 Dwight> a single chip combination of a number of 2901's in one
 Dwight> package. I think it had a 2910 as well. I forget if it was 4
 Dwight> or 16 2901's. Still, basically an extension of the 2900
 Dwight> bitslice family.

The impression I got is that it's a fairly distant relative of the
2901. It's a 16 bit processor. The 2901 lets you build a processor
of whatever width you like, and between the 2910 and other stuff you
can do pretty amazing execution stream control stuff. The 29116 is
more of a conventional microprocessor in the sense that it does what
the designers put into it, neither more nor less. Good if that's what
you want, because it's less work.

I once saw the UDA50 microcode -- which uses the 2901 and 2910.
Basically, the 2901 is an ALU, and the 2910 is the instruction fetch
and branch controller. In the UDA50, the microcode sources had two
instructions per line -- one the 2901 instruction, the other the 2910
instruction. There was a one cycle delay between the two, in other
words the 2910 on cycle N used condition codes set by the 2901s at
cycle N-1. This could produce weird looking code like this:

      clr foo ; bne bar

because the bne would react to the result of the previous line ALU
operation. It takes an unusually strange mind to cope with an
architecture like this. (No wonder Richie Lary liked it...)

             paul
Received on Mon Feb 28 2005 - 14:06:07 GMT