11 Mar
2015
11 Mar
'15
10:10 a.m.
On Wed, Mar 11, 2015 at 11:39 AM, Johnny Billquist <bqt at update.uu.se> wrote:
Ah yes, as you say. 256K.
> And if you
absolutely want to use 1M cards in an 11/70, ping me
256K, not 256M, but anyway... :-) There are several bits to it.
The MK11 only have 18 address pins on the backplane, which means it cannot
directly address larger boards.
Right. That's why you have to add wires when upgrading an older
11/750. Just putting in new boards isn't enough.
In addition, there are 16 card select pins. One for
each slot.
Sure.... makes sense.
At power on, the memory controller in the MK11 writes
to
all cards in parallel, in order to initialize the ECC bits on all cards.
Ah ha! That's the not-obvious trick.
This is not so hard. Essentially we're talking
about a OR of 4 lines, and a
4-2 encoder.
Sure. This part makes sense to reverse the "fan-out" for the per-slot select.
Important additional detail is that you *really* want
to make sure that when
all cards are select together, that is equivalent to the lowest addressed
cards.
This totally makes sense in light of the initialization phase.
This means that the ECC for 3/4 of a 1M card will not
be initialized.
Once you've come this far, you need to write a short program that will
initialize the ECC of the rest of your memory... You can turn off ECC
checking in the MK11 by changing a CSR register... on the memory bus,
and not on the Unibus. And the CSR address is in I/O space...
Thank you for a very good explanation of the pitfalls. So I take it
that refresh isn't an issue then?
I'm not likely to try this since I do have enough 256K boards for my
11/70s (and I have plenty of places to stick 1M boards in VAXen), but
it's interesting to see what it takes.
Thanks!
-ethan