Re: Double Buffer RK11-C

[Paul Koning]

> On Dec 6, 2016, at 2:56 PM, Fritz Mueller  wrote:
> 
> 
>> On Dec 6, 2016, at 7:51 AM, Noel Chiappa  wrote:
>> 
>> [data fetch] can't be off-loaded onto a separate interface unit, as it needs 
>> access to
>> register contents held in the CPU.
> 
> Yeah, it’s pretty interesting!  My guess would be that it was a separate 
> register/command oriented interface, sitting on the Unibus, and didn’t 
> actually interface directly with the 11/20 CPU?  Such an interface could 
> limit the instructions “fed” to the FPU to those accessing its internal 
> registers, etc.  But who knows? :-)

I don't know anything of a DEC product along those lines, but a college 
classmate of mine (Bill Black, Lawrence Univ. class of 1975) built a floating 
point coprocessor for our PDP11/20 that was a Unibus peripheral.  I helped with 
the software interface.  The device had 4 registers, two for source and two for 
second source and result.  They appeared at several different bus addresses; 
you'd select the operation to perform based on which address you used.  The 
device would start when the 4 source words had been loaded, then a read cycle 
of the result register would simply be held off until the operation was done 
(since it would complete well within the SSYNC timeout).

The implementation took, if I remember right, one hex-sized wire wrap board.

paul

Re: Double Buffer RK11-C

[Fritz Mueller]

> On Dec 6, 2016, at 7:51 AM, Noel Chiappa  wrote:
> 
> [data fetch] can't be off-loaded onto a separate interface unit, as it needs 
> access to
> register contents held in the CPU.

Yeah, it’s pretty interesting!  My guess would be that it was a separate 
register/command oriented interface, sitting on the Unibus, and didn’t actually 
interface directly with the 11/20 CPU?  Such an interface could limit the 
instructions “fed” to the FPU to those accessing its internal registers, etc.  
But who knows? :-)

I’ve gotten quite deep into the design of the FP11-B and associated KB11-A 
interfacing during my debug (which is how I noticed all the 11/20 refs in the 
docs, circuitry, and microcode), but I’m pretty ignorant of the 11/20 having 
never worked on one.

> I wouldn't be surprised if there's some microcode in the KB11 to support 
> those memory operations.

Yes, there certainly is — quite a bit of it actually.  The are F/CLASS branches 
off all three of the A, B, and C forks.

—FritzM.

Re: Double Buffer RK11-C

[Noel Chiappa]

> From: Glen Slick

> You should also hang out at vcfed forum in addition to hanging out here.

I barely have time to keep up with the things I'm already into (actually,
strike that - I _don't_ have time to keep up already :-)! So I'm going to
have to pass.


Interesting about the FP11 stuff, though. I wonder, however, if those 11/20
mentions are more aspirational, rather than something that was actually done,
and later dropped? I say this knowing a modest amount about how the KA11 CPU
in the 11/20 works, and a bit about how the FP11-B interfaces to the host CPU.

For one thing, it would take some modifications to the KA11 to recognize
floating point instructions, so that it can wake up the FP11-B to handle
them. But that turns out to be the easy part; it would take further mods to
the KA11 because the FP11-B expects the host CPU to do data fetches, etc on
its behalf. (See Figure 2-3 in the FP11-B MainManual.) This latter function
can't be off-loaded onto a separate interface unit, as it needs access to
register contents held in the CPU.

I don't have the time to look into this further, but if someone is interested
they'd need to study the FP11<->CPU interface, and in particular look at how
the KB11 supports memory operations requested by the FP11-B - I wouldn't be
suprised if there's some microcode in the KB11 to support those memory
operations.

Noel

Re: Double Buffer RK11-C

[Glen Slick]

On Mon, Dec 5, 2016 at 6:40 PM, Noel Chiappa  wrote:
> > From: Fritz Mueller
>
> > Also, clues about an 11/20 interface for the FP11-B that were noticed
> > recently.
>
> I don't recall this; more details, if possible? Thanks!
>
> Noel

You should also hang out at vcfed forum in addition to hanging out here.

http://www.vcfed.org/forum/showthread.php?55131-FP11-B-what-is-signal-FICC-EXT-ADD-L

Re: Double Buffer RK11-C

[Noel Chiappa]

> From: Fritz Mueller

> Also, clues about an 11/20 interface for the FP11-B that were noticed
> recently.

I don't recall this; more details, if possible? Thanks!

Noel

Re: Double Buffer RK11-C

[Noel Chiappa]

> look at the lower right line of lights on the panel: ... and three bits
> of Major State; now look at the RK11-C prints, connector B32:
> ... Postamble, Checksum, Data, Header, Preamble.
> ...
> One thing I have been wondering about is that "RK11-C" - that implies
> that there was a -B, etc. I wonder if this panel goes with one of them?

Well, now that I look at a few more things I'm pretty certain the panel in
that image goes with some currently-unknown RK11 predecessor to the RK11-C.

Note those 5 'state' lines/lights, and then look at the 'Major States' RK11-C
print (RK11-C-04, pg. 14 of the PDF version, RK11-C Enginering Drawings
Feb1971). In the upper left corner there are a row of 6 flops, each labeled
with one of those states (plus one for 'Idle'), arranged in a chain. So one
light for the output of each flop...

Now look at that display panel: 3 bits for 'Major State' - implying it is
binary coded - likely implemented with a counter?

Notice also the signal "COUNT MSR" ('Major State Register', I expect) - just
what you'd expect to see if the major state had previously been held in a
counter, not a string of flops.  Why go to all the trouble to synthesize that
signal (on the next page, RK11-C-05, "MSR Control") when you cou;d have used
the individual composing signals to clock each flop?

So my _guess_ is that in the previous version, they'd used a counter, but had
had some problems (perhaps it was a binary counter, not Gray code, and the
decoding into states was producing glitches), and had therefore switched to
the string of flops.

(This whole process makes me feel like a paleontologist, reconstructing some
unknown dinosaur from a fragment of one bone, using a lot of complex reasoning
from small clues contained therein! :-)


It would be most interesting to know if there are any signs anywhere of
predecessors to the RK11-C.

My suspicion is that they were produced in very small numbers - perhaps as
prototypes, only internal to DEC. (If they'd had problems with glitches in the
major state counter, they would not have wanted to release it as a product.)
Or if it was released as a product, perhaps they were all recalled and
replaced with RK11-C's because of the issues.

As evidence for this, I point to the Spare Module Handbook, which lists only
the RK11-C and -D - but _does_ list the KT11-B, a rara avis indeed. (More
dinosaur bones... :-) This argues that the predecessor did not exist in the
wild...

 Noel

Re: Double Buffer RK11-C

[Noel Chiappa]

> From: Fritz Mueller

> But then over at bisavers, I see this:

Yes, that's the panel I found the picture of in the RSTS-11 brochure a while
back:

  http://www.classiccmp.org/pipermail/cctalk/2016-November/029104.html

The picture in the brochure is not very very hi-res, but Al Kossow recently
located some original photos in the DEC section of the archive, and the image
you sent the URL for is one of them.

I've been puzzling over what this thing is.

It looks 'kinda-sorta' like an RK11-C panel (the registers it shows are sort
of the RK11 registers), but if you look at the print for the "RK11-C
Indicator Connectors: RK11-C-23" (page 34 of the RK11-C Engineering Drawings
PDF), you can see the pinout, and it doesn't match.

E.g. look at the lower right line of lights on the panel: 4 bits of Bit
Counter, a blank, 8 bits of Internal Word Counter, a blank, and three bits of
Major State; now look at the RK11-C prints, connector B32: 4 bits of Bit
Counter, 8 bits of Internal Word Counter, a blank, Postamble, Checksum, Data,
Header, Preamble.

Close, but different.

One thing I have been wondering about is that "RK11-C" - that implies that
there was a -B, etc. I wonder if this panel goes with one of them? (Or perhaps
it is a custom prototype?)

I have never been able to find out anything about an earlier version of the
RK11: the earliest Peripherals Handbook that I have is the 1972 Red/White
one, and it only talks about the -C. Also, the Spare Module Handbook (a
fantastic resource, it lists the boards in almost every PDP-8/10/11 option)
mentions the -C and -D, but no other ones.

However, given the example of the KT11-B, which was totally unknown until the
documentation for one showed up with the -11/20 in Arizona, thereby proving
that there _was_ a KT11-B before the KT11-C (the -11/45's MMU), I would guess
that there likely _was_ an RK11-B, and perhaps this panel goes with that (or
an earlier one).

Any further information would be most welcome.

Noel

Re: Double Buffer RK11-C

[Fritz Mueller]

So, the RK11-C disk controller page over at gunkies says “no [indicator panel] 
inlay for such has even been sighted…”. But then over at bisavers, I see this:

http://bitsavers.trailing-edge.com/pdf/dec/pdp11/pictures/1120_RSTS-11/1120_RK_Panel_RSTS-11.jpg
 

 

Re: Double Buffer RK11-C

[Noel Chiappa]

> I'm working on the prints now.

OK, done and uploaded:

  http://ana-3.lcs.mit.edu/~jnc/tech/pdp11/RK11-C-DB_EngrDrws_Dec72.pdf

I haven't had the time to pore over them to figure out exactly what the
changes do, but they add two buffer registers (ABUF and BBUF), so they
probably offer greater resilience to DMA contention on the UNIBUS.

I don't yet know if they are used for write as well as read (the RK11-C-DB
block diagram in the prints suggests not, as it doesn't show a path from any
RK registers to the ABUF, just from the UNIBUS - i.e. only usable on reads),
or if there are any user-visible programming changes (I suspect not).


For those with an RK11-C, these prints are somewhat clearer than the prints
for the 'basic' RK11-C which are online, so although there are a number of
changes (see:

  http://gunkies.org/wiki/RK11-C_disk_controller#Engineering_drawings

for info on which sheets have changes), use of these can help decipher some
of the hard-to-read pages of the 'basic' RK11-C drawings.

Also, the prints for the 'basic' RK11-C are missing a couple of pages:

  18 - Disk Cable and Termination
  19 - Bus "D" Drvrs and Rcvrs

which are present (albeit perhaps modified, for the second one) in this set.

Noel

Re: Double Buffer RK11-C

[Noel Chiappa]

> Check out the module utilization chart .. On a stock RK11-C, slots 1-8,
> rows C-D (the bottom two rows) are empty. ... (I'll be documenting the
> added Flip Chips in the Double-Buffered variant 'soon'.)

I've added a module chart for the double-buffered variant here:

  http://gunkies.org/wiki/RK11-C_disk_controller

I'm working on the prints now. (Any idea what I should call this thing?
There's already an "RK11-C_Engineering_Drawings".)

Noel

Re: Double Buffer RK11-C

[Noel Chiappa]

> From: Ethan Dicks

> I haven't even made an inventory of it. What would I look for to know?

Check out the module utilization chart, either in the RK11-C Engineering
Drawings, or here:

  http://gunkies.org/wiki/RK11_disk_controller

(at the top of the section "RK11-C Board chart/count tables"). On a stock
RK11-C, slots 1-8, rows C-D (the bottom two rows) are empty. In the "Double
Buffer" variant, they are full of Flip Chips.

(I'll be documenting the added Flip Chips in the Double-Buffered variant
'soon'.)

Noel

Re: Double Buffer RK11-C

[Ethan Dicks]

On Thu, Nov 17, 2016 at 10:09 PM, Noel Chiappa  wrote:
> > Subject: Does anyone actually have a KT11-B?
> > Date: Fri Sep 30 19:04:47 CDT 2016
>
> > the ones shown in the images show it to be (mostly) an RK11-C.
> > ...
> > I say "mostly" because there appear to be extra cards on the right hand
> > end; whether those are some sort of upgrade to the RK11-C, or whether
> > someone just stored spare Flip Chips out there, I have no idea.
>
> So this mystery has been solved (sort of).  According to some drawings I have,
> there is apparently something called a "Double Buffer Disk Control" variant of
> the RK11-C. I looked online, but there was nothing about it there, and the
> RK11-C Engineering Drawings at BitSavers don't cover this variant. Does anyone
> out there have one of these?

I have an RK11-C that I've never attempted to power on.  I haven't
even made an inventory of it.  What would I look for to know?

-ethan

Double Buffer RK11-C

[Noel Chiappa]

> Subject: Does anyone actually have a KT11-B?
> Date: Fri Sep 30 19:04:47 CDT 2016 

> the ones shown in the images show it to be (mostly) an RK11-C.
> ... 
> I say "mostly" because there appear to be extra cards on the right hand
> end; whether those are some sort of upgrade to the RK11-C, or whether
> someone just stored spare Flip Chips out there, I have no idea.

So this mystery has been solved (sort of).  According to some drawings I have,
there is apparently something called a "Double Buffer Disk Control" variant of
the RK11-C. I looked online, but there was nothing about it there, and the
RK11-C Engineering Drawings at BitSavers don't cover this variant. Does anyone
out there have one of these?

Noel