fpga implementations have been discussed before; while high speed fpgas 
exist, theyre very expensive.   I think a quasi-microcode implementation
holds more promise from both a speed and a cost perspective. Alphas can
be rather VAX-like, and theyre also in the spirit of keeping if DECish, 
and a couple of them can probably, with the right palcode, do the job of 
playing the part of a vax. a nexus interconnect, etc, from that on, with the
memory (perhaps with another alpha doing vax memory management for it) as 
a nexus itself as in the bigger vaxen, could probably be served well from 
one of the modern high perfromance schemes like sokolov mentioned his 
choice of. As for XMI, i think it's fanciful to aspire to get detailed enough
info to do a proper implementation, and the payoff would be nearly nil. 
I have just about zero XMI devices i really want to run unless im running
a vax 6000 with them, and even then, my vax4000 is faster and has a scsi
interface. PCI is gross, and i like the idea of VME. the new VME stuff is 
really bloody fast. I'd still like to see a fast scsi interface that 
talks MSCP, though.. 

It would be interesting to see if his project progresses, but i have to 
agree with thor, fpgas arent the way to go for a high speed vax. 


 On Tue, 26 Feb 2002, Thor 
Lancelot Simon wrote:

> Oh, yeah, sure, if you ignore everything about it that's wildly at odds
> with reality.  Care to tell me how he's going to make his FPGA with his
> new VAX core on it run at anywhere near the clock rates the later VAXen
> ran at?
> 
>