knudsen (02/22/83)
In his interesting speculations + help request the other day, Pournelle mentioned that the war between 8086, 68000, (& I guess Z8000) might be pre-empted by the "16032". What the heck is that? Is it made by National, or HP, or whom? Is it just a joke (look closely at that number)? If real, will it be able to overcome the software inertia/momentum that the other machines will have built up by that time? If not, can it run Smalltalk & fit in a briefcase? Maybe it's a bit-slice family -- build a 40-bit version for BASIC floating point, hey? mike k
guy (02/23/83)
The National Semiconductor 16032 is a 16/32 bit micro (like the 68000). It currently features a true demand-paged memory management unit (unlike that excuse for an MMU that Motorola provides). The machine is very VAX-like, except that the MMU has reference bits. It has what looks to be the nicest instruction set of the 16-bitters. There currently exist two ports of 4.1BSD to it, one by Human Computing Resources and one by NS and some other people. I don't know the available clock speeds offhand (4, 6, 8/10 MhZ I think), and I don't know whether production chips are ready yet (I suspect not). They are also working on an IEEE standard FPP (which will plug into the 16032, unlike the Motorola FPP which requires the 68020). It's a very nice chip in principle (architecturally, I think it is the best of all the 16-bit and 16/32-bit micros); however, it may be too little, too late, and it may never overtake the 68000. On the other hand, it does have a real demand-paged MMU... Guy Harris RLG Corporation ...!decvax!mcnc!rlgvax!guy
bernie (02/23/83)
The 16032 does in fact exist, and it is a *very* nice architecture. I have some information on it, but unfortunately non is machine-readable. An outfit called HCR (Human Computing Resources) in Toronto, Ontario has Unix ported to a 16032 system; I can't vouch for the product, never having used it myself, but the company seems sound enough (I have friends who have worked there). Also, a company out in California (SRI-TEK, if I remember right) has got a 16032 board coming out for the IBM PC. --Bernie Roehl ...decvax!watmath!watarts!bernie
eager (02/24/83)
The 16032 is indeed a National Semiconductor Corp. product. It was described in the April 24, 1980, issue of Electronics, page 123. The 16082 memory management unit is described in the current (Feb 24, 1983) issue of Electronics. Briefly, it is a 16 bit processor, well designed, high-level language oriented. It can work with a co-processor, if that appears. The chip is in production and boards using it have been out for a while. It was about a year later in getting to the market than its target. I'm not aware of a battle between the 8086 & 16000. If there is one, it is not clear to me that the 16000 will make any difference. It is very late to the market and has few supporting chips. Time will tell. -- Mike Eager amd70!eager
mdrutenberg (02/24/83)
The National 16032 is a real and @i(very) nice microprocessor. It is a joy to work with because it is so clean and regular. I understand it looks very much like a scaled down Vax, although I don't know enough about the Vax architecture to confirm this (it seems to be true though). -Mike
grunwald (02/24/83)
#R:ihnss:-145700:uiucdcs:10400046:000:493 uiucdcs!grunwald Feb 23 18:42:00 1983 If I am not mistaken, there have been articles on the Net about people using the 16032. I think someone mentioned that they had a C compiler, and then another article said something about BBC Computers (English I would guess) making a system using it. Additionally, I have seen ads for the "power tower" system (16032 based from National Semi) in MiniMicro Systems. Perhaps one of the people who work at National Semiconductor and who use the network will let us know some more abou it?
bcase (02/24/83)
#R:ihnss:-145700:uiucdcs:10400047:000:193 uiucdcs!bcase Feb 23 19:35:00 1983 Yes, you are correct, I have also seen some articles about the 16032 used in actual products; I had fogotten about that when I wrote the original response. Perhaps there is some potential....
guy (02/25/83)
The rumor I heard about the 16032 is that it was originally intended to *be* a VAX-11 lookalike, but that National Semi had just gotten sued by DEC for making a PDP-11 lookalike (and lost) and decided against it. Yes, it is quite like a VAX, Semitic bit/byte/word order and all. 8 32-bit general registers, plus the PC, static base (see below), (stack) frame pointer, stack pointers (user and kernel mode). It has VAX-like addressing modes, with either the 8 GPRs or the PC, Static Base, Frame Pointer, or Stack Pointer used as a pointer register. No auto-increment or decrement though. Also, it has *four* indexed modes (a la VAX); for some reason, the operand size is encoded in the index mode rather than being taken from the instruction. The addressing modes are less "regular" than the VAX, and seem more tailored to the use algorithmic languages would have for them. The instructions are the standard collection of two-operand highly-orthogonal instructions that any general register machine *should* have. No CRC16 and the like, and no full packed decimal. It has the extract/insert bit field instructions, and index checking/stepping instructions (like VAX-11 "index"). It has a similarly orthogonal set of instructions for the floating point coprocessor (IEEE standard single- and double-precision), and a set of instructions reserved for a "custom" coprocessor. The MMU is a standard demand-paged MMU (24-bit virtual address in, 25-bit physical address out) with 512-byte pages. The page table is itself paged. Furthermore, in a very un-VAX-like step, the page table has *REFERENCE BITS*! It has a large set of registers for program debugging; it can remember the addresses the last two branch/jump/call instructions came from, and can support two breakpoints - the break can occur on a read, a write, or an execute from that location. The data sheets that I am reading this from were handed out at UNICOM by National Semi. The chips are the NS16032 CPU, 16082 MMU, and 16081 FPU. Guy Harris RLG Corporation ...!decvax!mcnc!rlgvax!guy