Does it compute?

[Lobster]

This looks neat
http://kurobox.com/revolution/what.html

However unless someone is up for it we will stick with the x86 processor. AMD are developing low power X86 processors. No cooling. Smaller machines etc.

As a matter of interest what is involved in recompiling Puppy for another processor? Which one would be the best? What would the process entail?

[Sage]

Sadly, it's not just a case of technical excellence and scientific rigour. Motorola 6800 series legacy failed because Apple wanted to practice the same greed of leverage as M$. Everything they did/do costs more and clone second sources are not generally available (it used to be worse). Earlier, they insisted on using hugely expensive scsi HW for desktops - massive overkill. I remember spending weeks trying to replace some part or other for a friend in the US and failing because Apple wouldn't tell us how to set the ID, spin-up delay, termination parameters, etc. Even BSD's OSX has been proprietarised. Sooner or later punters will get to see through attractive packaging - won't they?!
AMD and Cyrix gained market share by being cheaper whilst still performing mundane tasks adequately! VIA is in the ascendant but is too expensive, presently - let's hope they learn the old lessons. Not everyone uses Linux, but a substantial number of PC users have now heard of it and know it is 'free'. We need to have the trade sell PCs, not windows boxes!
Some say IBM made an huge mistake in releasing IPR for the PC. That incorrect presumption is the big mistake that so many capitalists continue to make. It was purely bad management (as usual!) that caused the hiatus for Big Blue in the 90's. To quote a former CEO "What possible use could computers have outside a university laboratory". The PC has been a runaway success and continues to be a milch cow - more of it should have ended up in the IBM coffers.
It's the same stupid mistake that the RIAA, MPAA and the anit-pirating software people are making. (Suing impoverished students {and their parents!} is truly moronic). When Napster and Kazaa were running at peak, sales of legitimate music CDs boomed, as can be seen from the balance sheets of eg HMV. It's why M$ is, ultimately doomed to fail. Piracy strenghtens market penetration. Pirated products were never putative sales - folks accessing illegal copies haven't got any money!
The oldest lesson in the book, known to every street-corner dealer, - the first one is 'free'!! Profit comes from add-on services.
Klaus Knopper has become famous (not sure about rich?) because his principle product fills a need, is well-conceived, competent, supported and 'free'. One presumes he makes a tidy income from selling consultancy services? Good thinking Klaus!
Barry is certainly famous and might become rich if wished. [Probably, he's a happier man working on his pet and surviving. Riches just bring misery.]
Nobody really wants PPC, Apple, M$ - caught once, better informed next time! Too many bean counters, pen pushers and shareholders to support.
Stick with Linux, Puppy and x86 - you know it makes sense.

[Sage]

There you go, Lob! Guess we'll have another three hours to wait for the transAtlantic backlash! Sorry I'm not competent to address your questions. But I know a man who can.

[pythoncoder]

As a matter of interest what is involved in recompiling Puppy for another processor? Which one would be the best? What would the process entail?

The processor is only part of the story, you have to consider the entire machine architecture. In principle a change in the instruction set merely requires an appropriate C compiler, although even then there are complications with word lengths and big-endian or little-endian architectures. Also some processors support only memory-mapped I/O whereas others have a separate address space (with separate instructions) for accessing I/O devices. Then there are the complexities of hardware memory management and virtual memory; these are likely to be handled in different ways for non-Intel processors.

The architecture of the machine outside the processor presents its own challenges. Interrupt controllers, DMA controllers and such like will have their own unique register sets and methods of programming them. It's also likely that other I/O devices will use chips that differ from those used in the PC. I could go on, but I think you'll get the picture: unless a machine is specifically designed to conform to the PC architecture, it's likely to be a non-trivial task.

Unless you can acquire kernel source written for that platform...

The Kurobox's main virtue is that it is extremely cheap. If you're going into production of some kind of server device this is a major factor. However, for the amateur I think that mini-ITX is the way to go, on the grounds of PC compatibility. It's also available in a range of types offering different compromises of processor power/price/peripherals.

Pete

[Lobster]

Thanks Sage, Thanks Pete

Nathan has mentioned a possible project to port Puppy to older Apple hardware. However I would rather stick with x86 processors

[pythoncoder]

I'd strongly agree: stick with the PC architecture for Puppy.

For those into alternative hardware, there is the SLUG. A SLUG is a modified NSLU2 (network hard drive) that can run linux OS. This is probably the cheapest Linux embedded solution out there, although it's strictly command-line stuff. See

www.nslu2-linux.org/

Pete

[Sage]

A real tour de force, python. Good to read your complete exposition.