Subject: Re: Don't buy a vax, but the vax (was Re: RIP, VAX)
To: None <PORT-VAX@NETBSD.ORG>
From: Roger Ivie <IVIE@cc.usu.edu>
List: port-vax
Date: 08/28/1999 19:00:50 
> I also want, however, as ivementioned before, to see devices taking full 
> advantage of the power of the VAX architecture to do I/O the right way: 
> with devices acting as VAX nexi on a nexus interconnect, like the XMI and 
> BI VAXen were. (anyone who's seen a VAX 6000 knows the kind of awe you 
> get just looking at how these things talk to each other.. :) 

Sigh. I bet I'm the only one here who has ever used a DR-780. Nice
interface, and it influenced some subsequent work that I did. In a
subsequent interface, I had a Z80 sitting on the VAXBI looking up 
userspace addresses in the VAX page table. Fun, fun, fun!

Roger Ivie
ivie@cc.usu.edu
se we want the modules to be interchangable. And i do want the processor to be a separate module talking on a nexus interconnect! At least have bus adapters on this interconnect, if were not putting real devices for now. DEC already has wonderful nexus IO devices like SGEC, SHAC, CBQIC, and so on. Those are made for CDAL, but could be used as a base point or even used directly with a gate array or two doing some interfacing. isildur On Sat, 28 Aug 1999, Roger Ivie wrote: > The answer is obvious. You pick a fast processor (maybe a really hot > Alpha, or perhaps a fast PowerPC) and treat as a microengine for your > instruction set; i.e., you plunk it down with a ROM containing an > emulator that the chip starts executing out of reset. Then when the > commodity guys up the clock on their Alpha or PowerPC, voila! you've just > had a performance bump. I suspect a really hot processor could handle > (say) the Unibus in software, as well.