[pups] What is a 21-21858 chip?

[Carl Lowenstein]

> From owner-pups at minnie.cs.adfa.edu.au Sat Feb 17 18:54 PST 2001
> X-Authentication-Warning: triangle.cs.uofs.edu: bill owned process doing -bs
> Date: Sat, 17 Feb 2001 21:43:51 -0500 (EST)
> From: Bill Gunshannon <bill at cs.scranton.edu>
> To: pups at minnie.cs.adfa.edu.au
> Subject: Re: [pups] What is a 21-21858 chip?
> MIME-Version: 1.0
> 
> 
> Maybe I'm just misunderstanding what is being discussed here, but
> I think the difference is that without the FP11 all you have are
> 4 simple Floating Point Instructions.  FADD, FSUB, FMUL and FDIV.
> The FP11 adds a number of additional Instructions.  I have never
> had a machine with the FP11 (Hmmmm, wonder if my new 11/93 has it?)
> so I don't know them off the top of my head, but my Macro-11 book
> is near at hand if anybody wants me to look it up.

You're missing something.  The FADD FSUB FMUL FDIV was first used in
the 11/40 FIS (optional) floating-point instruction set.  It was later
implemented as an optional chip add-on for the LSI-11 and LSI-11/2.
Single-precision only, and stack oriented.

The 11/45 Floating-point Processor had about 30 instructions and had both
single-precision and double-precision modes, and a bunch of registers.
Typical instruction mnemonics include ADDF SUBF MULF DIVF MODF and
the same with last character changed to D for Double.

The LSI-11/23 optional FP chip followed the 11/45 FPP style.  The 11/73
CPU chip implemented the 11/45 FPP in microcode, and the 11/73 FPP
add-on implemented the 11/45 FPP in much faster microcode.

    carl