On Mon, 10 Oct 2005 01:41:07 +0100 Jules Richardson <julesrichardsonuk at yahoo.co.uk> wrote: Xenix runs fine in the 512KB of the Altos 586, and supports multiple users in that memory.

Tony Duell wrote: Yep, you're right (about there being empty sockets on both versions of the board). I suppose it's logical it'd be an MMU chip given that there's no other connectors to the board (e.g. it's not for an optional serial chip or whatever). I don't know offhand the number of pins on the MMU support chip for the 32016, but I'm guessing 40 pins as that's what I remember most of the larger chips in the WCW MG-1 having :) No, I haven't either. And interesting speculation as it perhaps implies that Acorn were thinking of a UNIX workstation very early on (and may also accout for the rumours about Xenix being available; I bet if Acorn were thinking along those lines it ended up in various pre-launch magazine articles) No, nor me. Apart from those which are broken and don't run anything :) cheers Jules

Tony Duell wrote: Funnily enough, mention of that socket just cropped up on the BBC list, and someone remembers their father working on one of these boards with a floating point unit fitted into that socket via a small daughtercard. They're willing to admit that they may be misremembering though! Would that be a possibility though, or is it definitely an MMU socket? They're going to mention it to their dad, but suspect that he's long since forgotten details too. (Incidentally this was at Oxford uni - which is the first time I've heard that they had any 32016 units, although I suppose it's not surprising) Oh, and I was also thinking about the Xenix on the ACW rumour. It's remotely possible that someone managed to get Xenix running on the 80286 coprocessor in the ABC 3xx machines I suppose, and that's where this rumour of ACW's running Xenix started (after all, the machines look identical externally). It'd solve the porting to a different CPU issue at least, but presumably would be no small feat to get it to do I/O across the TUBE. Still, slightly more plausible than an ACW version... I could see Acorn doing early press releases about the ABC range and saying "oh yeah, we've got Xenix in the works for one of these" - and of course to most people the ACW and the ABC equate to the same system as the Z80 version's largely forgotten and the 80286 version never existed outside the labs (far as I've been able to tell). Hence over time things got corrupted into people thinking there was a 32016 Unix variant... cheers Jules