Date   : Sat, 04 Mar 2006 18:04:16 +0000
From   : Jules Richardson <julesrichardsonuk@...>
Subject: Re: Warning: Sad case on list!

David Hunt wrote:
> Hehehehe, another Beeb nut works its way out of the woodwork.
> 
> I can't see why the 32016 SP on eBay would be duff, the hardware is almost
> bullet proof. 

Agreed.

 > Although spares are going to be difficult, I've seen the Math
> Co-Pro chip on its own going for $100.

Yep, that's why I've got a spare 4MB board here - plus I grabbed a very old 
clunky laser printer the other week just because it has a 32016 in it.


> The big 4Mb board on eBay recently
> would be fun since it runs at a decent clock speed, 10MHz vs. 6MHz of the SP
> and has 4Mb of breathing space vs. 1Mb.

Hmmm, porting 42nix to it would be interesting ;)

Actually, the few fragments of documentation about the Xenix port to the ACW 
are interesting, as it seems that the whole "all I/O done through the TUBE" 
aspect was really causing Acorn headaches.

 >  Mind you, I'm not convinced by the
> 32016 as a CPU, it is hilariously slow

Hmm, one of those cases where it depends on usage I suppose; ISTR for some 
things it could outperform the competition.


> In theory (I believe) if we could extract the BIOS ROM from the
> 6502/Z80/32016/80186/80286/68000 2nd Processors 

I *almost* have all of those on disk here - just missing the Casper 68K ones I 
think (although I have the Torch 68K ones).

I'm still looking for genuine '286 ones, if they ever actually existed - the 
'286 powered ABC machines were pressed into service by Acorn for Master 512 
development, so they all ended up with Master 512 ROMs (which is what my board 
has). I'm not even sure if Acorn ever even got to the point of fitting the 
'286 copro with its own ROM, or if it was such a stillborn project that they 
immediately started life as Master 512 dev boxes...

Unfortunately there aren't enough surviving '286 copros around to check!

 > we could use a C/C++
> emulation library for the corresponding processor and implement the Tube
> interface and then run it on Sprow's ARM 2nd Processor. 

Hmmm, I don't know how such things work (never written an emulator before in 
my life), but I know the Torch 68K boards do a bit of funky messing around 
with memory protection via a PAL chip - so it's not just as simple as 
emulating the CPU. (I'm not sure if the Torch 68K ROM code expects the SCSI 
hardware to be present too - I've got a feeling that they might)

cheers

J.