On 9/3/07, Petter Urkedal <[EMAIL PROTECTED]> wrote:
> I think we can afford to use 8 bits for encoding registers.  That leaves
> us 8 bits of the instruction word for other uses, which is sufficient in
> the current version.  Further, we don't need to split evenly between
> register kinds, so maybe something like
>
>     q0..q31   previous frame registers r0..r31
>     r0..r63   current frame registers
>     g0..g127  global registers
>     s0..s31   IO-registers
>
> would be more useful.  We should keep in mind, though, that we no longer
> have computed addresses, so we can't e.g. do table lookups.  That is,
> unless we have a spare memory block which we connect to two of the
> IO-registers.  (We have not quite walked in circles.)

I think these are interesting ideas, but I just don't think we need
this complexity for something whose entire program file is only 512
words.  If we decide later to develop a completely free computer with
our own CPU design, we're going to have to address these things, along
with virtual memory.

-- 
Timothy Normand Miller
http://www.cse.ohio-state.edu/~millerti
Open Graphics Project
_______________________________________________
Open-graphics mailing list
[email protected]
http://lists.duskglow.com/mailman/listinfo/open-graphics
List service provided by Duskglow Consulting, LLC (www.duskglow.com)

Reply via email to