There's a fairly extensive writeup by the designer, Monte Dalrymple, in the current (January 2006, #186) issue of Circuit Cellar.
True, still no on-chip memory, but there's an expanded register set and the external bus width can be set to 16 bits to improve bandwidth. There's a funky bit of code to get the thing booted before it knows what the external bus width actually is.
.... for which they've added a bunch of support for 32-bit operations.
-- Dave Tweed
For the business side of things, you can read the last couple of quarterly filings from DIGI (who bought Rabbit/ZWorld in May). For the $50M price tag, about $30M of that was for "intellectual property", which seems to mean RabbitSys and the Rabbit 4000. That's a good chunk of money - apparantly DIGI thinks the new features are compelling.
They will likely do just fine selling new and/or improved Rabbit 4000 core modules to the same folks who buy Rabbit 2000 and 3000 core modules but need more grunt.
Kelly
Yes, a sizable chunk, but Monte's website
- one hopes DIGI realised this, so the IP would seem to be their Variant C, and all the Rabbit/Z World modules/libraries [PCB designs will be classified as IP ]
True, and maybe they will do a stacked-die variant, that has bulk CPU/FLASH/RAM ?
-jg
There's a free assembler here:
Look for as85.zip. I don't know how good it may be.
If/when Atmel offers ASICs that way I imagine Rabbit'll order them that way ;)
Kelly
later. I believe the 8085
that may just be old
Well, your memory jibes with mine - see you at the old folks home...
Bob
That doesn't mean that there weren't S-100 boards that used the 8085 -- the SpaceByte comes to mind.
Norm
Didnt mean there were not others, or that they were UNimprtant to someone, just that the Z80 was the most common/important CPU for CP/M.
CP/M would run on most any similar processor with a little coaxing.
8080, 8085, 8086.. even the 8051 could do it in theory..But by far the z80 was the king of the hill ..
too bad IBM chose dos.. might have been a different world today.
Amen
Are you sure? I was deeply involved with CP/M in the late 70's - early 80's and I don't recall seeing it run on anything but an 8080/Z80.
CP/M-86 would indeed run on an 8086 and run well, but it was a totally different animal than CP/M
1.4 or 2.2, which I presume we were talking about.
I do agree that the other versions were 'different', as they did use different machine code, but the OS calls were the same at least. I wasnt so much talking a particular version, as the 'concept' of CP/M.
i dont have examples at this stage of the game, its been too long. But i do remember seeing it on chips in the same sort of loose family. Im sure they never were of much use due to the binary incompatiblites.
No, not the 8051. The 8051 was quite different from the 8080 compatible processors. The 8080, 8085 and Z-80 (and a few compatibles) all executed the 8080 instruction set as a base, but all (except the
8080, of course) had various additional instructions. The Z-80 extensions were quite large. The 8051 is for embedded applications and is a totally different ISA.
yes, it also used a harvard arch natively.. but it could be made to run like a more traditional cpu too, with a few tricks.
While it might not be an easy weekend project, cpm could be done..
Well, sure, but it would have been a complete rewrite, just like the
8086 and 68K versions of CP/M. And running a more traditional OS and applications with a more traditional view of memory would have been painful to say the least. Just think about how painful references to memory past the first 128/256 bytes on an 8051 are. Everything has to be done as indirect references with MOVX through DPTR. Admittedly with a unified address space you could also do indexed indirect loads view DPTR or PC, but that's not going to help much.Could be done, and reasonable to do are often different things.
The 8051 is really not targeted at this market and the architecture reflects it.
No.
CP/M 2.2, the most common flavor, was written in 8080 assembly language.
Assuming you bought the source from Digital Research, you'd still have to hand-translate it to another instruction set. I don't recall any automated conversion tools for the 8051 in that day.
I worked at Digital Microsystems where we purchased the source for BDOS and ended up doing a complete rewrite because of the unsupportable quality of the code. And we paid something like $5000 for it.
It would have been great fun hand- translating that code to an 8051.
Thanks for that link Jim. This brings back some memories! I still have an old Panasonic JD-840 lying around in the attick (securing the house by it's weight). I also maintain an access-control system I designed years ago, running a stripped-down CP/M kernel so I could use Pascal/MT+ from MT MicroSYSTEMS to compile the code. This compiler could generate rommable code and I ran it in a CP/M emulator (ZRUN) on DOS. I still keep a Windows 95 installed on a laptop to run this, since 95 is the last OS that supports FCB's.....
Meindert
ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.