Honeywall mainframe CPU front panel ID?

Dave Wade dave.g4ugm at gmail.com
Wed Jul 26 10:53:09 CDT 2017



> -----Original Message-----
> From: Dave Wade [mailto:dave.g4ugm at gmail.com]
> Sent: 26 July 2017 16:34
> To: 'Noel Chiappa' <jnc at mercury.lcs.mit.edu>; 'General Discussion:
On-Topic
> and Off-Topic Posts' <cctalk at classiccmp.org>
> Subject: RE: Honeywall mainframe CPU front panel ID?
> 
> 
> 
> > -----Original Message-----
> > From: cctalk [mailto:cctalk-bounces at classiccmp.org] On Behalf Of Noel
> > Chiappa via cctalk
> > Sent: 26 July 2017 15:38
> > To: cctalk at classiccmp.org
> > Cc: jnc at mercury.lcs.mit.edu
> > Subject: Honeywall mainframe CPU front panel ID?
> >
> > So, I've been collecting images of 'Multics' 'front' panels from
> > around
> the
> > Internet, intending to do a gallery.
> >
> > (I should explain that, in common with mainframes of that era, a
> > Multics system had a variety of different kinds of boxes - CPUs,
> > memories, etc -
> but
> > also others, intended to support the multi-CPU 'utility' concept. It
> > was possible to take, say, a running 3-CPU system, split off a CPU,
> > bring that
> up as
> > a separate system, then later bring that down, and add it back to the
> running
> > system! This was actually done at the MIT site, to allow development
> > work
> in
> > the evenings on the OS software.)
> >
> > The Multicians site has a nice picture of a Multics system with the
> > some
> of
> > the panels swung open (they're actually 'diagnostic' panels, so would
> > normally be swung shut):
> >
> >   http://www.multicians.org/mulimg/h6180-doors-open-big.jpg
> >
> > The CPU is the one in the center (the panel on the left is an IOM,
> > 'I/O Multiplexor', one of the other kinds of box).
> >
> >
> > So, anyway,I had this large collection of pictures, and asked: Tom Van
> Vleck,
> > the maintainer of the Multicians Web site what the other (non-CPU)
> > panels on offer might be, and his reaction was (roughly) 'some of the
> > CPU panels there might not be Multics CPU panels'.
> >
> > (Honeywell had an entire line, the
> >
> >   https://en.wikipedia.org/wiki/Honeywell_6000_series
> >
> > but most models in that line ran an OS called GECOS (later GCOS), not
> > Multics. So possibly those CPU 'front' panels are from some other 6000
> series
> > CPU.)
> >
> > His reasoning was that they don't have the Appending Unit sections: to
> > explain this, Multics used an extra box (the Appending Unit), inserted
> > between the CPU and the memory, to implement the paging and
> > segmentation of Multics, and most 6000-series CPUs did not have this.
> >
> > If you look at this image of what is probably the Multics CPU panel
> > now at
> the
> > LCM:
> >
> >   http://ana-3.lcs.mit.edu/~jnc/tech/multics/jpg/_1020903.jpg
> >
> > it has an Appending Unit section at the top. (BTW, are there any
> > pictures online of LCM panel? All I could find was the video, which is
> > admittedly
> > ultra-cool.) See the "APU Scroll" section (first full-width section),
> > for
> the
> > Appending Unit, at the top in this detailed shot:
> >
> >   http://ana-3.lcs.mit.edu/~jnc/tech/multics/jpg/_1020899.jpg
> >
> > It's not an extra panel: the CPU panel on a Multics machine, although
> > the same overall size, has a different configuration, with the APU
sections.
> >
> >
> > However, the suspect CPU panels don't have those sections; see an
> > image of one
> > here:
> >
> >   http://ana-3.lcs.mit.edu/~jnc/tech/multics/jpg/multics_panel.jpg
> >
> > with detailed images here:
> >
> >   http://ana-3.lcs.mit.edu/~jnc/tech/multics/jpg/multics_panel_cu1.jpg
> >   http://ana-3.lcs.mit.edu/~jnc/tech/multics/jpg/multics_panel_cu2.jpg
> >   http://ana-3.lcs.mit.edu/~jnc/tech/multics/jpg/multics_panel_cu3.jpg
> >
> > Which is not _definitive_ that they aren't from a Multics machine, but
> > it certainly raises a big question mark. So, the question is, 'are
> > they
> Multics
> > panels, just for some reason without the APU section, or what'?
> >
> > So maybe these are from some other Honeywell Series 6000 CPU? If so,
> > does anyone knows which Honeywell 6000 series machine (it pretty much
> > has to be from one of them) they are from?
> >
> >      Noel
> 
> From what I remember, and it wasn't often you opened the panels on the
> L66 boxes, the systems all had almost identical configuration panels.
> All CPU's were upgradable on site to any other model. There wasn't really
> any difference between the models, the differences in performance were
> got by "marketing tweaks".
> So the bigger models had cache, and if you had two store boxes the memory
> could be interleaved.  There was also a cache consistency feature.
> 
> Later models also had virtual memory which I think used the MULTICs
> hardware....
> 
> .. so whilst its possible to say a panel is not from a multics box, I
don't think its
> possible to say exactly which model it came from, and indeed as the CPU
was
> upgradable the same panel could have been on multiple models.
> 
> When I worked on one belonging to NERC, UK, it was upgraded from a single
> CPU L66/60 to a DPS300 and I am pretty sure none of these doors were not
> changed...
> 
> Dave
> 

Actually looking at this manual:-

http://bitsavers.trailing-edge.com/pdf/honeywell/dps-8/58009853_DPS8_46_70_R
eference_Man_Sep82.pdf

these are from the original hardware GE600/6000/L66/DPS300 machines. The
DPS8 had a redesigned panel...
.. page 2-25 (page 40 in my PDF viewer)

Dave



More information about the cctech mailing list