On Feb 14 2005, 17:24, Eric Smith wrote:
> Pete wrote:
> > early versions of the KDF11-A had 18-bit ODT while later versions,
> > and all KDF11-Bs had 22-bit ODT.
>
> I wrote:
> > Are you sure that there was a difference in the ODT? I thought the
> > ODT was the same in all F11 chipsets, and the difference was purely
> > a matter of whether the KTF11 MMU was installed, and whether the
> > module wired the top four address lines to the bus.
>
> Pete wrote:
> > No, I'm *fairly* sure there was a difference in the ODT. Perhaps
> > Sellam can tell us for a change; if he has an 11/23 (KDF11-A) with
> > 18-bit ODT (his MXV21 responds to 777170) and also an 11/23-plus
> > (KDF11-B) with 22-bit ODT (his MXV21 responds to 17777170 but not
to
> > 777170) he can tell.
>
> Wouldn't that be explained just as easily if the ODT was identical,
> the logic that generates BBS7 ignores the top four address ines on
the
> 18-bit-only processor module?
The effect would be the same. Unless it's the case that some cards are
18-bit ODT but can handle 22-bit addresses through the MMU.
> Aha! Section 5.8 of the manual says:
> During normal address cycles, BBS7 L is asserted if BSIO H is
> asserted by the F11 data chip or the MMU chip. During ODT
address
> cycles, the ODT logic peformans an ODT relocation cycle to clear
> BBS7 L if either ODTA17(1)H or ODTA16(1)H is clear.
> If this documentation is correct, ODT always uses only 18-bit
addresses.
> I don't have a KDF11-B to check.
It probably is correct; my memory was probably being more "random" than
"access". Moreover, I've just spotted a comment in my original QBus
course notes about ODT being 16, 18, or 22-bit and "so it doesn't
necessarily match the memory addressing capability". There's also a
note that an early KDF11-A "can be upgraded from 18-bit addressing to
22-bit but it's not easy". Given that was written around 1984 when it
would have cost real money for a DEC upgrade, and written by a hardware
engineer who did component level repairs on the things, it probably
means "don't bother trying".
So it would seem I was wrong. It's not 18-bit/22-bit ODT that is the
difference between KDF11-A, but 18-bit/22-bit memory access.
> I'm not sure if my KDF11-A is an 18-bit or a 22-bit version. Anybody
> know what etch and circuit revisions introduced 22-bit capability?
Ah, well, that one I am sure about :-) Rev.A was the 18-bit version
and Rev.C was the 22-bit. You can tell the difference at a glance by
looking for W18, which is a red wire across a resistor in the upper
centre of the PCB on a Rev.C.
--
Pete Peter Turnbull
Network Manager
University of York
Received on Tue Feb 15 2005 - 02:26:08 GMT