M7264 Troubleshooting

[allison]

On 06/08/2019 10:37 PM, Noel Chiappa via cctech wrote:
>     > From: Allison
> 
>     > ODT for the two systems are very different. .. KDF-11 the ODT is part
>     > of the higher level code. The larger cards (11/23 and 23+) boot to
>     > resident (ep)rom.
> 
> Ah, no. (Well, the KDF11 CPU's can boot to EPROM, which in the -11/23+ can be
> on the CPU card; the -11/23 is a dual card and has no functionality on the CPU
> card except the CPU.)

Yes that is what the para quoted above implies.  The resence of Eprom
and serial are unique to the F11 based and later J11 quad width cards.

All of the Qbus processor cards could boot to Eprom (TEV11) just like
the Unibus 11s.

And yes the KDF11 cpu does rely on higher level code inside the
microcode its how that code was structured [and made to fit].


Also there are three cards for the KDF11, one dual width, and two
quad width.  The 11/23 early did not boot all devices and had
different eproms (board level difference) than the later 11/23+
hence the + [and different part designations in the Mseries]
designation.  Back then FS noted that when requesting replacement
boards and history requires it.

> 
> The ODT in the KDF11's (and KDJ11's) is, just like in the LSI-11's,
> microcode, not macro-code. From the 1982 'microcomputers and memories'
> handbook, pg. 161 (in Chapter 7, "Octal Debugging Technique (Microcode
> ODT)"):

Save for the CPU and microcode are entirely different.  ODT as a
function is defined to do certain things how its done is vaguely
similar at best.  While implemented in ucode how its done and
depnendancies are very unique to the each (again cp1600 and F11)

>   "The console emulator Octal Debugging Technique (ODT)is a portion of
>   the processor microcode ... The console ODT implemented on the LSI-11/23,
>   PDP-11/23 and PDP-11/23-PLUS is identical."

Yes and unlike many here I have CPUs of all three form factors
operational.  However ODT on the dual width CPU does require a
serial card as there is no way to talk to it without.  That is an
important difference especially if jumpered for ODT.  Actually my
"11" series Qbus collection includes all of the CPUs from the LSI11
though J11 based. And the bus versions are greater from the LSI11
early (and H11) though the later ones with PMI some specific to
devices like the RL11 controller board set.

However LSI-11/23 whatever that is, typo? So I will say the CP1600
processor of LSI-11 and the F11 (KDF-11)processor have major differences
in how ODT works internally and on the bus and the code that does that
are very different.  To the user with a terminal its not very visible
but to the service person (or someone assembling a system) it makes a
difference.

ODT had a specification and if you reffered to it that inside DEC it
was not clear if it was microcode only that what the user/field service
saw behaved in a very specific way.  When applied to a specific
processor it had deeper meaning.  Some of that was factory test
related.  That Specification was an evolved thing as LSI-11 (cp1600)
lead to additions and minor changes that were important to Field
service and manufacturing.

> and on pg. 154:
> 
>   "Unlike the LSI-11 and LSI-11/2, the LSI-11/23 does not enter console
>   ODT upon occurrence of a double bus error"

Glad to see someone reads the books.  There are other differences on
power up and run states.  Like the ram and console dependencies

But all the M7264 posts were boiled down to the problem of not
reading and understanding that ODT for that version of the CPU
had dependencies.  As well as the evolution of Qbus family of PDP11
CPUs and their low level and bus level idiosyncrasies.

Most never refer to the LSI-11/23 that way to mean KDF-11 CPU its was
a documentation error that propagated in educational services and lead
to errors.  LSI-11 (CP1600) was the first generation Qbus and later was
F11 or J11 (or the T11).  The difference was not always small as there
were subtle instruction set differences and diagnostic impacts.  That
always had me during my yeas at DEC going which one are you talking
about, as every thing had at least three names (never minding numbers)
and one was usually ambiguous or a nonspecific family name.

> From which I think is quite clear that the KDF11's have microcode ODT.

It's not a question only the implications of how they did it.  That
they did it using using ucode is too road a generalization and misses
implementation details.

Its those details that get you when assembling systems and forgetting
to include functions that are required if not used.  That goes back to
how DEC supported their systems when computers no longer had front
panels.  It was a field service requirement so they were not trying
to wake a brick with nothing.


Allison

>       Noel
>