I just googled for comparisons and posted links here.


Meh, if there was only a good benchmark of PandaBoard's CPU and an Atom dual core. I could make an easier decision...


Actually, it is around that area. 600-800 KB/s is what I see.

Sustained throughput? Piping from what, Flash (or an SD card)? That would be extremely limited in terms of storage, though; my point is that using it as a NAS media server, by somehow adding more storage, would make it completely unable to sustain that throughput.

For an iPod Touch, obviously the flash chip inside. Apple hates SD cards for some reason.

USB is something of a mystery. I guess it should be fast, since ARM is a mobile platform, and so any communication should be speedy. Filesystems can make it slow - that is, the NTFS rw usually munches CPU. But I don't think it would slow it down that much.

It doesn't necessarily follow that a mobile platform would be good at communication, although I see your logic. It wouldn't make much sense for it to be optimized past the bandwidth that most cell networks can support, for example, if you're dealing with a cellular platform. I'd actually be fairly interested to see what kind of streaming sustained bandwidth one of these boards could pipe out, but I'll admit I don't have high hopes.

I'm pretty much set on purchasing a PandaBoard (decent specs, $174). I'll be sure to tell you the results of my testing! Of course, I need to magically get $174 out of thin air...

Anyway, here's another question: what if I just want a mini wifi box, in which you can interface with it via mini USB or link port? What would you suggest I use? (And yes, there's something interesting I'm planning to do - you can guess. )

"Link port"? What kind of a link port does the PandaBoard hae?

No no, the PandaBoard will be used for my tiny server. The wifi box is something that will be a future part of C2I to have true portability (like your gCn, you gotta have it tethered to the computer). 2X1GHz ARM CPUs are a bit overkill for that.

Well, good luck surpassing all the other coders to date that haven't found sufficient processing and memory power to interface wifi modules.

Looks like you misunderstood.
The wifi stuff is in the box, and the calculator doesn't handle wifi connections. The only parts it controls is entering any keys/pass, and choosing a wifi network to connect to. The rest will be handled by the modified C2I client inside the box, powered obviously by Linux.

Unless the Z80 gains the capability to handle wifi communication/encryption, I'm pretty sure wifi is all external.

With that said, what specs/dev board should I use?

Welll, I guess with that in mind something like a Gumstix might be your best option, although they're notoriously hard to interface with something like a calculator. To be honest, though, this sounds suspiciously like instead of connecting the calculator to a computer to get it on a network, connecting it to a smaller computer...

These are notoriously expensive too... cheapest is $129?? Are there any other cheaper options?

And you're kinda right - this portion of C2I is connecting the calc to a tiny computer with the C2I client installed. However, the regular C2I is just the regular calc to computer tether for internet.

Yeah, but for $129 you're getting a full tiny computer running Linux, so it's not really that expensive if you think about it. It does seem a bit silly to have a calculator tethered to a device that costs twice as much as the calculator itself, though.

True indeed. I have seen cheaper ones though, and I'm pretty sure Linux runs just about anywhere.

Any other dev boards that you may suggest?

When by "just about anywhere" you mean "something with a processor of an ARM, MIPS, or x86 type at least, a few MB of RAM, and a few dozen MB of Flash", then yes, it does run "just about anywhere". I can't say I have any other dev boards to suggest; what are you narrowed down to at this point, assuming that you're still set on creating this project as planned despite our notes and warnings?

Yup - that. MIPS is an alternative, but I like to use ARM since it's a well known platform.

I'm still set to this project - and I still think you have no idea what I'm trying to do. The calculator does not handle any wifi communication; the linux mini box will. Simple C2I communication back and forth will be the key to success. This project won't start that fast though - C2I itself needs to be finished! (I'll toss gCn onto the box as well when you are done with that)

That said, I'm aiming for cheap, cheap, cheap. $10-30 max for total cost of board, excluding any LCDs. Must needs are Linux support (I don't feel like coding an OS or doing Wifi comm. by hand), wifi support (can be external chip, if needed), and USB.

No no, I was warning you against the whole media server on an ARM dev board project; isn't that what this thread was originally about? I thought the calculator-related discussion was just a sidetrack from the original points.

Oh... in that case, I'm already set on buying a Pandaboard. I still feel that it would be very effective as a server, but it won't be the best for sure.

Then I look forward to some tests and benchmarks once you get it and try it out.

Not necessarily true, Kerm. ARM hz count more than x86 hz, because of Instructions per Cycle. ARM IPC counts are usually 2-3(it depends on the instruction), with a few outliers, where x86 is higher.

Yes, but each instruction can do more in a CISC chip than a RISC chip.