PDP-11/83 w/FPU?

[Paul Koning]

> On Apr 17, 2019, at 4:36 PM, Ethan Dicks via cctalk <cctalk at classiccmp.org> wrote:
> 
> On Wed, Apr 17, 2019 at 2:48 PM W2HX via cctalk <cctalk at classiccmp.org> wrote:
>> ?Hi friends. I am putting together a PDP-11 set up. i have a few CPUs available to me, a 11/23+, an 11/73 and I also have available to me an 11/83, M8190-AE .
>> 
>> I would like to try to run as many different OS's as may interest me, including some unixes as possible (bsd...etc).  My question is, are there any OSes that need the floating point option?
> 
> I know old UNIX (v2) needs a KE11-A or KE11-B (a Unibus peripheral
> integer math option), which really wasn't something people
> bought/installed in an 11/34 or later.  v5, v6, and v7 UNIX shouldn't
> require any sort of math hardware.

I think that was typically called "EAE" (extended arithmetic element), a Unibus peripheral that implemented integer mul/div and maybe a few other odds and ends.  RSTS V4 had optional support for that.  It only applies to 11/20 and 11/05 since all the other machines have the relevant instructions built into the CPU.  (Later versions of RSTS required that.)

> Simple OSes like RT-11 do not require it and the The usual DEC
> timesharing OSes (RSX-11, RSTS/E...) won't require it,, though you may
> encounter a FORTRAN application that was compiled to require it.

In the case of RSTS, you could build your BASIC-PLUS either for FPU or not.  If you tried to run the FPU version it would of course fail badly without that option, but you could then swap in the other one and applications would run fine.

One curious detail: in the very early days DEC had a floating point software library that uses three words per value: 2 word mantissa, 1 word exponent.  The source code for that popped up recently in a collection of TOPS-10 DECtape images, I remember writing an email to the list identifying it along with various other items.  RSTS V3 (and I assume before) used that float library.  Then in V4 they switched to the new standard, your choice of 2 or 4 word float which matches the FPU formats.  Apparently they pulled that change without any means to convert existing binary data files; probably there weren't enough to worry.

	paul