Ragge said:
>> I've had the dubious honour of adding a vax 3520 to my little VAX
>> museum. I want to put netbsd on it, but I can't figure out from the
>> instructions whether it is supported. 
>> 
>If your machine has Q-bus and disks on the bus (not on-board ctlr)
>then it should work fine. But it's two CPUs in that machine, eh?
>You will only be able to use one of them.

Sorry, the 3520 is a bit more complicated than that.

The 3520 is a dual-processor CVAX machine built around a proprietary
write-back cache bus called MBUS (not to be confused with Sun's MBUS).
Each module interfaces to the MBUS through a standard chip called the
FBIC (Firefox Bus Interface Chip (Firefox was the internal codename for the 
machine)) which has a number of options and must be properly configured for 
each module.

A minimal configuration of a 3520 would be a CPU card (containing two CPUs, 
each with its own cache and FBIC), an I/O card (DSSI, Ethernet, and four
serial ports), and a memory card. The machines were also shipped with a 
graphics card set which is allegedly pretty hot (it was designed during the 
DEC and E&S get-together), but I've never run it through its paces, so I 
can't say.

QBus is optional on the machine. There are two different possible QBus 
adapters: FTAM and FQAM. FTAM is built by gluing a CQBIC on a bord next to 
the FBIC. Because of the incredible fun involved in the write-back cache 
bus, FTAM is only guaranteed to support the TQK70 tape controller (the FTAM 
can generate QBus latencies of up to 20 microseconds). The FQAM is built 
from lots of jelly beans with a microcoded controller; it is much slower, 
but guarantees that it can meet QBus maximum latency timing. Hence, FQAM 
should be ablee to support any QBus module which can put up with a slow 
QBus (I've run RQDX3, RL02, and RX02 on an FQAM).

NetBSD would need work to support the MBUS and the FBIC chip in order to 
support the 3520. MBUS info was tightly held; the only third-party 
interface I am aware of to MBUS was the MasPar's interface to their 
massively parallel box (MasPar was founded by a former DEC vice president; 
in addition to info on MBUS, they got a good deal on buying the machines).

The 3520 never sold well, so I don't expect a lot of them to show up.

Roger Ivie
ivie@cc.usu.edu