Re: Flame Linus to a crisp!

[John Bradford]

> > With open hardware designs, there would be no problem with
> > documentation not being available to write drivers.
> 
> See below...
> 
> > Incidently, using the Transmeta CPUs, is it not possible for the user
> > to replace the controlling software with their own code?  I.E. not
> > bother with X86 compatibility at all, but effectively design your own
> > CPU?  Couldn't we make the first Lin-PU this way?
> 
> In theory; in practice we have no access to documentation.  See above...

I'm now stuck in a mail reading loop, with the see above and see
belows :-)

> That makes Transmeta part of the _old_ industry :)
> 
> I believe present Transmeta CPUs are quite specialised for x86
> behaviour (memory model etc.) anyway.  When you're running on a CPU
> like that, there's probably little to be gained from changing to a
> different front-end instruction set.
> 
> Special tricks like non-cache-ping-ponging locks and faster interrupt
> handling might improve performance, but probably require a change of
> the hardware to implement.

Shame.  I guess it wouldn't really have got us any closer to an open
hardware design anyway, it just seemed like a nice hack :-).

John.