Acorn Business Computer

[yevrowl]

There is a line of PCs that are quite diverse in terms of functionality and even architecture — Acorn Business Computers.

Interested in whether there are emulators for them, as well as interpreters and compilers for the BASIC programming language.

 

[Pernod]

The ABC range are basically just BBC B+ machines with either a Z80, 32016, or 80286 co-processor bundled in a big box, they were intended to run CP/M, Panos, and DOS respectively. There's nothing special about them and very few were actually produced, only the Acorn Cambridge Workstation which includes the 32016 co-processor was actually marketed.

The ABC range are emulated in MAME. They all contain the standard BBC BASIC interpreter, just like a standard BBC B+, but the ACW was bundled with FORTRAN, PASCAL, C, and LISP languages.

 

[silcreval]

I have a couple of these systems and they are a lot of fun. Putting everything in one case saves some deskspace, and the systems were surprisingly well built. Panos was useful enough, and the compiler set was pretty good for the time. I've seen a few apps over the years, Kermit was created for the ABC and some other packages.

 

[Cruxinc]

I have a couple of these systems and they are a lot of fun. Putting everything in one case saves some deskspace, and the systems were surprisingly well built. Panos was useful enough, and the compiler set was pretty good for the time. I've seen a few apps over the years, Kermit was created for the ABC and some other packages.

Wow! Thats quite something to have, they are super rare. Would you be able to share any photos/specs of your machines?

 

[silcreval]

Wow! Thats quite something to have, they are super rare. Would you be able to share any photos/specs of your machines?

Yes would be happy to. They are rare but the 32016 co-processor was surprisingly fast I think - and would have been decent in its day. I have the Panos software as well in shrink wrap, possibly not too many unopened boxes these days.

 

[Cruxinc]

Yes would be happy to. They are rare but the 32016 co-processor was surprisingly fast I think - and would have been decent in its day. I have the Panos software as well in shrink wrap, possibly not too many unopened boxes these days.

That’s great, thank you!
The 32016 is amazing, it’s my second favorite processor architecture, and my Cambridge co pro is one of my favourite items, by a long way, and use it near daily. ABC even better-one of my dream computers- what do you use it for?

 

[stepleton]

the 32016 co-processor was surprisingly fast I think - and would have been decent in its day.

It would be interesting to encounter a 32016 being speedy. The only setting I've encountered one in is a Whitechapel MG-1 workstation, which feels quite slow. The OS's compiler is pcc though and I've been told that it doesn't put out particularly efficient code.

 

[Eudimorphodon]

It would be interesting to encounter a 32016 being speedy. The only setting I've encountered one in is a Whitechapel MG-1 workstation, which feels quite slow.

It looks like the MG-1 uses the 32082 MMU, which makes memory accesses take a *huge* hit if you go outside the small translation cache.

Outside of that I guess you'd just need to calibrate your expectations accordingly. On paper the 32016 looks like it should be at least a little faster than the 68000 because it has a fully 32 bit ALU (IE, it's "actually" a 32 bit CPU with a 16 bit bus, verses the 68000 being a 16 bit CPU that "acts like" a 32 bit one), but on the flip side it has a multiplexed data/address bus, so... it's probably just a wash in the real world.

 

[Istarian]

The 68000 is something of a hybrid, really. It's not exactly a 16-bit CPU nor is it quite a 32-bit CPU. --- It has 32-bit registers, 32-bit arithmetic instructions, and a 24-bit address bus, but the internal/external data buses as well as the ALU are all 16-bit.

It would be a 16-bit CPU if we defined these things solely by the width of the data path (esp. the internal one), but the hardware itself is capable of working with 32-bit data...

So logically it should perform similarly to a 32-bit CPU running at a slower clock speed.

 

[Eudimorphodon]

So logically it should perform similarly to a 32-bit CPU running at a slower clock speed.

Sure. Mainly what I was pointing out is that what's happening "under the covers", so to speak, can vary between different CPUs that externally (or according to their API/instruction set) act similarly. My impression is that National Semiconductor tried to harp on a particular difference (ALU width) to demonstrate their CPU's "superiority" to the older 68000, but the CPU's failure in real-world tests (and ultimate market failure) points out that theoretical advantages might not actually translate into real-world performance.

(FWIW, it probably wasn't really the cycle-time performance of the 32016 that killed the chances of it becoming the mainstream choice over the 68000, it was more a combination of it being late to market and, initially at least, really buggy, but that it didn't perform much differently than the chip that'd already been available for a few years certainly didn't help. I think a lot of people today actually kind of misunderstand/over-estimate the role that "performance" had in the personal computer world for the first few years. Even if the NS32016 had been *substantially* faster the Motorola CPU specifically incorporated some features that made it an attractive upward migration path from 8-bit machines, which I suspect played a major part in it becoming the CPU of choice for the majority of "inexpensive" first-gen 32 bit computers like the Macintosh/Amiga/etc.)

 

[Istarian]

FWIW the Wikipedia page says it was initially called the '16032' (as opposed to '32016').

So they might have been better off sticking to that moniker and working to improve performance...

It also claims that in addition to being buggy that the CPU could not run at it's rated speed.

Motorola might have won mostly by being first to the market and not having any serious flaws.