On Sun, Oct 07, 2001 at 04:20:08PM +0200, Anders Magnusson wrote:
> > Ragge,
> > 
> > Long, long ago, you wrote:
> > 
> >  The VAX 8350 system is probably the slowest multiprocessor system              
> >  that NetBSD will ever support; each CPU does about 2VUPS (~2MIPS) :-)          
> > 
> > No 785? hm, I know the other CPU doesn't have access to I/O, but is this a
> > Problem? Or is the memory non-shared?
> > 
> I assume you mean the 782, 785 is just a hotted-up 780 :-)

ah, I guess, yes...
 
> The 782 is in principle two 780's, each of them with separate IO devices,
> separate memory etc but with some shared memory (in a separate box) to
> make the two machines to be able to communicate with each other. The
> shared memory is intended only to send messages between the two machines,
> it's slow and normally very little memory. Its intended usage is something
> like the SYSVSHM that communicate between processes.
> 
> So, the 782 machine is just two "normal" single-CPU machines that can talk
> to each other faster than over network. I know Mach was ported to that type
> of VAXen but I think that was only for testing purposes to see if the Mach
> concept worked on it. It wouldn't be realistic to try to port NetBSD to it.

Didn't VMS spawn user-mode processes to the other CPU on that type of machines?
I've read - in my far-away youth - a pretty detailed description of that, so
maybe I'm thinking of the wrong machine...

Regards,
	-is