16 Feb
2015
16 Feb
'15
12:23 a.m.
Ah. I need to count better. There are *two* MKA11 per CPU, thus each only
*needs* just a 1 Mw configuration due to address-space limitations.
On Mon, Feb 16, 2015 at 3:05 AM, Paul Birkel <pbirkel at gmail.com> wrote:
Notice the two unused connectors on the M8162 there in
the middle slot
(#15).
As this looks like an operational configuration, I infer that those two
connectors are used for ... nothing? Test points (through series
resistors)?
I see (how can one miss!) the address and data ribbon-cables. The
remaining white ribbon-cables must be for the box controllers. So, that
looks like a 1Mw configuration. Maybe 1.5Mw if, as you suggest, you're
careful?
Agree about the missing documentation :-<.
Thanks for the better pic-pointer; knew that I can seen those in the past
but couldn't recall where (and Google didn't help).
On Sun, Feb 15, 2015 at 8:56 PM, Eric Smith <spacewar at gmail.com> wrote:
It appears that the module utilization of the
MKA11 is significantly
different than the MK11, so it must have a much different backplane.
The photo of the MKA11 only shows eight storage array modules, in
slots 9-12 and 18-21, though slots 5-8 and 22-25 might also be
available for storage array modules. The memory bus cabling is
obviously quite cumbersome, and adding storage modules in slots 22-25
looks like it could exacerbate that problem.The MK11 supports 16
storage array modules in slots 2-9 and 18-25. The module assignments
for the non-storage modules also appear to be different.
It's a shame that the KB11-Cm and MKA11 field maintenance print sets
aren't anywhere to be found. While the available 11/74 documentation
offers a description of how the KB11-Cm differs from the KB11-C
(mostly interlocked ASRB and cache bypass), it's a shame that we can't
study the actual details of how those changes were implemented.