and which do you think would be best for this?

reason I ask is that I have plans to build a device that will need a Parallel port that would have thousands of hard wired pins if not millions and nothing exists today that could handle that. so the idea is if a brand new style of computing were devised requiring all levels of programming to be done from scratch who do you think would be the best candidate to make a user friendly OS for it?

remember BRAD NEW like using trianary or quadrinary as opposed to binary with multi sensitive transistors that would have 4 or more pin outs instead of just high and low but be able to notice variables like ultra low and ultra high with maybe a high mid and low mid range of electron counts.

another alternative would be to use current processors but build a motherboard that had 50 or more sockets that allowed the OS to make parallel use of them (these sorts of machines have been built but I don't think any of them take advantage of current processor tech that is widely available)

the first one will probably have a whole bank of Parallel port style 25 pin D-sub ports and a bundle of cables running from machine to machine but in the future this would change and maybe even be magnetic based to allow easier connection and disconnection. (just get it near and it pulls the connector in the right placement)and of course micro sized to allow for smaller foot print

Also any ideas on how to get the specs for the machine to the community of programmers would be appreciated with this discussion as I will monitor the posts for those sorts of links so I can add them to my bookmarks I carry on my thumb drive:D

Not sure you need a really new computer with a new OS created from nothing.
The standard PowerPC systems already have standard parallel ports on standard PCI slots. And the high-range PowerPC's have several hundreds of PCI slots (if you really need so many of them). And these machines run standard Unix operating system, so nothing new to start from scratch.
Now you only need to know what you really want to do, and how to do it using standard Unix programming, C-programming if you want fast response times.
You know, creating new machines is a lot of costly effort, if you can do your thing with standard machines, even expensive ones it's more reasonable.
And what I love with PowerPC is that you use the same processor and the same operating systems from the small pizza-box on your deks, until the huge machine with 64 cpu's on the same motherboard and several Gig's main memory. So, you can start small for testing, knowing that scaling to the final size is only a matter of having money enough for buying standard devices. So you know where you come from, you easily do prototyping, and you know where you are going at which cost.

What's sad is that a lot of advancements that are available on other more superior architectures like PowerPC aren't being tapped because of Intel and AMD being sticklers to the x86 architecture (and also Microsoft's fault). What got me really pissed off was when Apple swapped over to Intel, this just damaged the diversity and the industry as a whole since IBM/PA Semi are pumping out amazing chips that would've landed in Apple machines and increased the competition. Sure, I know IBM was plagued with manufacturing issues, but Intel chips are plagued with running too hot. Apple rushed their decision on it. I love Apple, but there are certainly flaws in their decision.

xboxrulz

I agree with you, xboxrulz. I too was sad in a way when apple decided to abandon the PowerPC architecture. I believe a lot of Mini-ITX boards use a Via C3 architecture, however, and I am hearing that a new Risc machine may be made.

morphious69, It looks like you need a IO subsystem. Not really a pc for what you've mentioned. If you need multiple input level, then you might need to use ADC(analog-digital-converter) for multiple level detection. There's many variance to select from, including but not limited to, sampling speed (Hz to GHz), resolution (level, from 8bit(256 levels) to 24bit(16million levels) or more). So, it sounds more appropriate to be using an embedded board with enough IO for your purpose. You can also link multiple boards to increase the number of IO. If you need further processing capability, then you can feed the data to PC, possibly via USB, Firewire or PCI and PCIe. Those for PCI and PCIe are call data acquisition card. They comes in many chooses of configuration as well. I think that's more practical than turning the good old LPT(Parallel Port) into something that it's not.