[comp.arch] separate integer and float regi

aglew@urbsdc.Urbana.Gould.COM (08/17/88)

>... what were the tradeoffs that went into the decision to make the
>floating point and integer registers the same?

[I didn't see the original post, just the quote in a reply]

Another reason for making floating point and integer registers
the same: you can use the mantissa multiplier of your FPU to
do integer multiplication. Rather than having a true multiplier
in your floating point unit, and multiply step instructions
in your integer unit, as most RISC systems with separate 
integer and floating point units, and register sets, do.