Any systems using the NS 32xxx CPU?

[Chuck(G)]

I just scrapped an old laser printer that used a National Semi 32016 CPU. I was surprised to find it, as the previous model of the printer used a 12MHz 68000.

At any rate, I remember the NS32xxx CPUs as being very late in production, but with a very advanced instruction set.

Does anyone own a personal computer that uses one of these chips? (32016, 32032, etc.) If so, what software was available for it?

 

[carlsson]

Did you check Wikipedia?
http://en.wikipedia.org/wiki/NS320xx

I know nothing about those particular systems, but it would seem the second processor in a BBC Micro might be the most familiar use of the NS32XXX. As you probably have read, Atari at the beginning considered building their 16-bit series on the National Semiconductor CPU, but quickly switched to Motorola 68000 before the ST got finalized.

 

[Chuck(G)]

The problem I have with the Wiki article is that it doesn't appear to be accurate. I've never heard of or seen a BBC Micro that used the 32016--to my knowledge, most of the models I've seen used the 6502, or, in the case of the BBC Master, the compatible 65SC12 The Acorn box used a special RISC processor based on the ARM architecture.

It could be that the NS32016 was available as an add-in/on board for these systems. Does anyone own such a board? In any case, that hardly counts, as ISTR that there was also an add-in board for the IBM PC using the NS32016 that never went anywhere.

 

[cosam]

I've never heard of or seen a BBC Micro that used the 32016
...
It could be that the NS32016 was available as an add-in/on board for these systems.

Kind of like an accelerator I guess - you could get "second processors" or "co-processors" which could do the heavy lifting whilst the 6502 was left to handle I/O. Whether you can consider a machine with a 32016 second processor an actual 32016-based machine is, I guess, debatable. That said, I'm pretty sure the Acorn Cambridge Workstation used a 32016 as its main processor from the get-go.

 

[carlsson]

Yes, there has been messages on the bbc-micro mailing list quite recently about restoring 32016 co-processors or at least the separate power supply. Certainly other co-processors may have been more common even for the Beeb. It is a bit like asking which computers use an early Power PC processor. While you can get an add-on card for the Amiga line of computers, only a limited number of programs support it.

But as you wrote, if the NS was a bit late to arrive, chances are that most computer manufacturers already had decided on either of the competitors. In that case it is even more interesting if anyone at all would bother to 1) make hardware add-ons and 2) write software for those. It really must have had an impressive instruction set and other strengths to still remain a viable upgrade option for some manufacturers.

 

[Chuck(G)]

I'm trying to pin down a date by recreating a mental picture of discussions (the company moved around that time) we had with various manufacturer's reps. I'd like to say it was 1980 at the latest. We had gotten an engineering sample of the Motorola 68000 and were looking it over (that big 64 pin DIP was murder on board real estate!) and had gotten some (or were about to get) very early silicon of the Intel 80186.

Since we were buying our "glue" logic from National, they sent over a couple of guys to make a pitch for their new 32-bit processors. The paper descriptions blew me away--paged virtual memory, module descriptor blocks, two stack pointers. Naturally, my first question was "when can you start sampling?". There were some looks exchanged, and then the response "we're hoping for 11 months from now". So every few months, I'd call them and get a "not yet".

Eventually we went with the 80186/80286 combination. But the IBM PC would shortly make its mark and put many operations out of business by the time the 32016 finally hit the streets.

I would have loved to have used it, however. The instruction set was elegant. I sometimes wonder if the PC didn't kill off a lot of innovation in processor design. After all, the 8086 was supposed to be a stopgap product until the iA4432 came out.

 

[gcoville]

In about 1986, I was able to play with the DB32000 evaluation board from National Semiconductor. It had two sets of eproms. One was called TDS which stood for Tiny Development System. It was for standalone use. You could connect up a dumb terminal and have access to pretty standard resident monitor. It also included a rudimentary text editor, assembler, and debugger.

The other set of eproms was for host-mode development. At our university, we had been given the VAX/VMS version of the development tools. In that situation, one of the configurations was to have the DB32000 sit between the VAX and the VTxxx terminal. As I recall, the downloading function would send a magic code which would put the DB32000 into capture mode. All the rest of the time characters were passed through to the terminal.

I think there was a separate mode to connect the DB32000 to its own serial port on the host computer, if you had one.

One of the benefits of the DB32000 is that you could try out all three variants of the CPU. There was the square quad socket for the 32032, and a more familiar 48 pin dip socket for the 32016 or 32008.

The board was designed for 10Mhz operation, but at the time I don't think they had 10Mhz CPU silicon. So it had a 6Mhz CPU, 10Mhz support chips, but was strapped for 6Mhz operation (clock speed, wait states, etc.)

My university was given 10 of these boards. But because the microprocessors lab had just invested in a small army of SDK-86 boards, it was pretty clear to me that these DB32000 boards would never be used. I never returned the one I'd been given to evaluate, and I still have it today! :sarcasm:

Around that time NatSemi also was selling a design evaluation kit which came in a small suitcase-style cardboard box. It contained databooks for the 32000 series, assembler syntax, TDS, product brochures, etc. It also contained a CPU, MMU, FPU, TCU, Interrupt Controller, and TDS in eproms. That kit came in 32032 and 32016 versions. I did buy one (the 32016 version) and wire-wrapped a simple system together. I left out the MMU and interrupt controller. Just last year, I dug out and played with my DB32000 and old wire-wrapped 32k board. They both still worked!

An interesting tidbit. There was still support in GNU assembler for 32k up until a few releases ago. The PC532 project had kept 32k going well into the 90s, and the support existed in the GNU tools for quite a bit after that, but I think has now been removed.

 

[Chuck(G)]

The chip on this board is an NS32CG16V-15 in a 68-pin PLCC. What I gather from the datasheet is that it's basically a 15MHz 32016 with a blitter and some additional graphics instructions. Most of the rest of the board is some big gate arrays, some mask ROM and a bunch of EPROM--and a bunch of SIP DRAM (256Kx4).

But it might be fun to fool with the NS32CG16--it doesn't look as if it would take a lot of support chips to get it going.

The curious thing is that the previous model of this printer used a 12MHz 68000.

 

[BlueofRainbow]

Ns32cg16?

Ns32cg16?

Hi,

Which laser printer did you dismantled and hence found the NS32CG16?

I've been looking for a working NS32K system for quite a number of years.....finding good links yet none leading to someone willing to part from one. Maybe I should be looking for laser printers?

Acorn, actually having worked with the NS32K for the Cambridge WorkStation designed their ARM Processor as an antithesis of this processor.

Interestingly, the elegance and symmetry of the instruction set is what attracted Prof. Wirth (ETH) for the design of the Ceres-1, -2, and -3 workstations used in their graduate and undergraduate computer laboratories. The book "Project Oberon" describes the software aspect of these workstations (Operating System, Language, and full source code). M.Eng. and Ph.D. Theses describe the hardware aspect.

I'm not sure if this was one of the first portable "Unix" systems ever but the Symetric /375 (i.e. half a VAX-750 in a smaller package) was based on the NS32K.

One of the co-processor board based on the NS32K (the DSI-32) was used as the hardware base for an urban planning tool for quite a number of years.

From a few old references, the NS-32008 was supposed to be able to run 8080/8085 object code natively......this might have been marketing words only.

Two more oddities:

i) A radiation hardened version was designed and fabricated for space missions. I think Scandia 3300?

ii) The last generation of the series NS-32764 (a.k.a. "Swordfish") never made it to market.....although some of the designers were given key-chains embedding their dies (self-destructed?)

P.S. gcoville - don't be surprise to receive one day a pm from me!