GMP on Pentium 2

[Kevin Ryde]

Torbjorn Granlund <tg at swox.com> writes:
>
> How about sharing some code between k7 and p6?

Oh, well, only that there's supposed to be a reason for why things run
the way they do, and divining it might even lead to 2.0 for p6 :-).
(2.0 is adcl throughput).

> We just found k7 code to be way superior, do we now
> need to invent something completely different for p6?

I might link it across for now.  Have to remember to undo it if k7
gets new code itself (less unrolling).