Ultrix V4.2 sources et al...

From: Michael Sokolov <msokolov_at_ivan.Harhan.ORG>
Date: Tue Mar 9 19:01:25 2004

Antonio Carlini <arcarlini_at_iee.org> wrote:

> The fastest you can get on Qbus is the DSV11.

I know, I have one, and I used it for my Inet connection when I lived in Dallas
and had a 64 kbit/s DDS line. It's rated for up to 256 kbit/s with one channel
active or 64 kbit/s with both channels active.

> That can almost
> handle 1Mb/s (it fails when fed a constant stream of minimal
> length HDLC frames so it was rated at 256kb/s - the next
> lower useful speed).

Do you have any more details about DSV11 operation above the rated 256 kbit/s?
What exactly is the real limit and what are the limiting factors? I have the
technical description (EK-DSV11-TD), but it isn't absolutely clear.

> The DSB32 is also
> not an option if you have a VAX 7000

I'm not using a VAX 7000 (and don't consider it a real VAX thanks to LSB and
whatever else they broke in NVAX+ when they fit it into an Alpha pinout), but
I'm curious why. I mean, how is DSB32 special? I recently searched for DSB32
on Google and came upon a posting by you saying the same thing, there you also
said it had something to do with the way its firmware locked memory. What
exactly is wrong with it? Does it strictly follow the VAXBI spec or was it
given an exception? Does it use IRCI/UWMCI pairs exactly as the spec says or is
it broken? Or is the VAX 7000 kludge the one at fault here in that the memory
is not presented to VAXBI as prescribed by chapter 8 of the VAXBI spec (the
Must Respond Set for Memory Slave), and DSB32 just happens to be the only
device that uses everything that the spec says is allowed? Or will all VAXBI
devices that use BVP (interlocked queues) not work on VAX 7000 because its VAXBI
interface to memory (through XMI and LSB) does not support IRCI and UWMCI?
That would be DSB32, AIE/AIO (DEBNx TK50/TK70/Ethernet), all BI-to-CI adapters,
DWBUA if some UNIBUS device accesses memory with DATIP-DATO (which turn into
IRCI/UWMCI), what else? You said DMB32 should work, I guess it isn't using
interlocked queues. But then again IRCI and UWMCI are the in the Must Respond
Set for memory nodes (section 8.2.2), so it would just mean that VAX 7000 is
not a real VAX as I've been saying all along.

> I suspect that for T1 speeds you'll be stuck using an
> Alpha or possibly a DECnis,

Well, neither Alpha nor DECnis firmware run 4.3BSD-Quasijarus, so those are not
my options (I love my 4.3BSD-Quasijarus based router and won't give it up), but
there is one more option: build a new sync serial card! We can even build both
Q22-bus and VAXBI versions.

Received on Tue Mar 09 2004 - 19:01:25 GMT

This archive was generated by hypermail 2.3.0 : Fri Oct 10 2014 - 23:37:04 BST