<   Hmm. This applies to complete MicroVAX and VAXserver boxes, right? Wha
<about CPUs sold by themselves. Suppose I had a 3500 and the KA650 died, o
<suppose I had a 3100 and the KA41 died. I could buy a replacement CPU,
<right? Would the price be different depending on the setting of the
<multiuser/single-user bit in the EPROM? Extending this idea further, it
<seems to me that it was possible to build one's own VAX from parts the wa
<people often build their IBM PC-compatibles. Think of it, one could buy
<separate CPUs, disk and tape drives, controllers boards, etc., right? Sur
<these parts were intended as replacement parts or upgrades for DEC-sold
<boxes, but anyone could buy them, right? The only thing that I don't kno
<how to get separately is the enclosure frame. But theoretically it can
<break too, can't it?

In all cases yes, or at least at one time you could not that would be 
unlikely and only things like boards and power supplies would be 
available.  Also you can get every piece at a price exceeding that of 
the closest system to what you are building.

< Maybe these were also available as replacement parts
<If they were, one could build his/her own VAX from parts, and it would b
<neither a MicroVAX nor a VAXstation nor a VAXserver, nor would it have a
<model number... Ideal for free OSes... It's also interesting how would th

Meaningless conclusion.

< KA650 and KA42/41 are certainly different enough to have completely
<different XSID "bitouts". They have completely different architectures,

You however confuse the _implementation_  such as Q-bus vs bounded with 
_archetecture_ which is still VAX.

They have SID as part of diagnostic strategy and also to help the OS 
configure.  It's pointless to be loading q-bus specific drivers in a non 
q-bus box.  Also there are low level implications of using a bus over 
direct interconnect involving performance like the presence of UBAs
which funnel 32 bit access down to 16 bit.  In the case of ka650 and ka42
it is the same CPU chip but what your talking about is implementation.
The closest thing to this would be around the time (3-4years ago) that 
Pentiums started really showing up and it could be on a board with 
ISA16/vesa or a different one with PCI and while it could litterally be 
the same Pentium chip in both cases the implmentation differences are 
there and in some cases significant.

<> separate video RAM.
<   Isn't it EK-VTTAA-TM? As for the video RAM, I think I know which chip
<it is in. I don't have a board that's not in an assembled box, so I can'
<point them out precisely, but the two chips that I'm suspecting are in th

Video ram is not one chip.

<far) both say that the host adapter SCSI ID is 6 and never even hint at a
<possibility of it being different. I have also played with one of my KA42
<It has a SCSI/SCSI mass storage controller daughterboard, and the host
<adapter SCSI ID is 6 on both buses. I have looked through all console
<commands, but I couldn't find anything that can change this.

Correct, as a host it's meaningless and really there to aid diagnostics. 
Also the bus probes may use it. It can be programatically changed by what 
ever code you care to execute to perform that.  However in the roms it's 
6.

<world has it set to 7. That post has kept me wondering since then, and
<since I currently have neither a KA41 to play with nor any MV3100 docs t

Its 6 in my 3100m76.  SCSI ID id in the host is only meaninful if the 
bus is multimaster. I've never seen the microvaxen SCSI bus use that way.
I've configured z80 systems (SCSI) as multimaster so the bus would serve 
in the same fashon as eithernet and the SCSI ID would be the analog of 
eitherent address.  However, the scsi as used in microvaxen is strictly 
master/slave with the disk/tapes being slaves.  This would be an 
interesting way to connect VAXen as it's fast even with PIO. Of course 
it's limited to SCSI distance restrictions but in a small setting it's
a useful interconnect.

Allison