Jeremy Fitzhardinge writes:
> Neil Russell bubbles:
> > The decision now is whether we go for it now, using PALs, or wait
> > 6 months and use what's available then. If we wait, there could be
> > a real possibility of using the R4200 and getting some real
> > performance. Given the speed that this project is going, it could
> > well be 6 months before we are ready anyhow, in which case we
> > should design now with the release date set some time around X-mas.
> > Comments?
I'd say go for it. The R4x00 would make the motherboard much
more attractive. I agree that chances are we ain't gonna finish
our motherboard before that anyway.
> Getting any kind of OS up is going to take at least a couple of months
> to get a simple framework up, and more to get a usable system (at a
> rough guess). It seems important to get a prototype system up quickly
> for doing software on, even if the first saleable thing is not until
> next year. I don't think much can be done without hardware, tho the
> assembler code can be rewritten and the kernel can be got compilable,
> given a reasonable spec of the hardware.
Lately I've been considering setting up a software simulator (like
they did in the famous smart frame buffer paper).
Using gcc+binutils to build a cross compiler and a tweeked version
of spim, we could headstart things like porting libc.
As we approch final specs more hardware could be emulated by
spim. I might have limited access to a mips machine, so I'll
start setting up a cross compiler.