You remember correctly: the 603 was never intended for multiprocessing. It was amazing it worked as well as it did on the BeBox, but AFAIK it was the only 603-based computer ever to try. The G3, the 603's more-or-less direct follow-on, doesn't support it either. From the BeBox FAQ: "We use the Motorola 105 'Eagle' support chip, which allows for either one processor and an L2 cache or two processors without L2 caches. So, our two-processor PowerPC BeBox has no provision for an L2 cache."
Aha! I am mildly surprised I remembered that correctly as I hadn't thought about it in yeeears.
Thanks for the info :)
Is there any memoir, book, whatever detailing out decisions on the hardware? Like why did they go with the 603? etc?
I really wish there was some deep-dive book on the early history. I can't (well, couldnt. Havent looked in a while) find much info on the original hobbit based system.
The initial BeBox prototypes used AT&T Hobbit processors and DSP’s. When AT&T dropped the Hobbit, they chose PowerPC over a risc system from HP. At that point in time, all these non x86 systems seemed to be gathering industry support. Then WinTel exploded into the market hungry for cheap vanilla systems.
IIRC, there was a big load of details on everything to do with the BeBox and all the decisions made and even a timeline and an interview with Joe Palmer on BeBox.nu. You can probably still read it via Archive.org. I wonder what happened to Andrew Lampert... he vanished off the web. I really miss BeBox.nu and lost touch with a bunch of people when it vanished.
Cale Lewis/Dlazlo - if you read this or if anyone knows him please chime in below ;-) He was like the best guy on those forums. Miss that guy.
Semi-related, I keep wondering how long we'll keep scaling hardware coherency out. I'd like to some 128 core+ chips that act like 2+ disconnected systems, just with high speed fabric.
