Picked up
from Slashdot, this just blows my mind. Some company, apparently French, has announced the
Jackito... A PDA (well, they want you to call it a 'TDA,' because they've given thought to stylus-free control) based around two impressively weak CPUs (a 20MHz M16C, and a TI all-in-one DSP), and some sort of Spartan FPGA with inobvious specifications. (One page seems to reveal a single
10MHz clock for everything.)
Claimed hardware specs are here, and a slightly goofy
block diagram over here. The
'dedicated automata' in the FPGA are listed here, and aren't much more exciting than the functions of a dedicated chipset... but it's hard to say what the flexibility to re-blow them will allow.
---
This is a really interesting approach to a low-power design, and could, perhaps, promise one h*ll of a different approach to software development. That benefit is muted a bit by the decision to make the standard SDK some sort of 'clip-on' to MSVC... which will allow for rapid development, but probably insulate coders from the sheer insanity of the architecture. (Without knowing more about it, I assume this mode of development hinges on how well the compiler can 'optimize' routines into the FPGA, and I'm going to guess they haven't actually left themselves a lot of spare capacity in it. The erstwhile
MyLinux used one simply for reasons of convenience, after all, and it takes two -- possibly larger, possibly not -- 'just' to implement a CommodoreOne.)
One thing's for certain -- right now, it looks like it can't handle JPEGs, let alone a browser... and while the FPGA *might* allow some miracles (at the least, performance on par with iBrowse on a classic 1200), you can get a
lot more conventional horsepower for the $600 being asked.
---
As such, I'm personally going to be ambivalent about it. On the one hand, this is a much more 'interesting' design than a Palm, for the "No, really, it's an organizer, stop thinking of it as a palmtop!" crowd, and I continue looking forward to the day when
something PDA-shaped hits the $20 point, with the flexibility to replace my remote controls, wall plates, wireless thermometer, dedicated-proprietary GPS mapper (well, if only I could afford one of those), etc. If this proves more likely to develop 'down' into that, more power to it. On the other hand, embedded CPUs keep getting more powerful, standards keep getting more standard, we're already well-past the performance of an A1000 or an early Sun 'workstation' in the handheld space, and maybe I just wish someone would come out with a single conventional platform with a full assortment of 'sane' real-computer features (a USB host controller, soft-loaded OS, and documented bootloader, for a start) instead of finding new and different ways to make crippled address-book replacements. After all, you're already going to own a (>Mac Classic-powered) cellphone if you care so much about that.
Whatever the case, this is one platform that int
ent might need a fair rewrite... or at least some careful thinking... to support.
---
Edit: I may as well
reference this link directly, too, where the pain of being a startup once again comes up as ever. The quote, emphasis mine:
"We will reduce this price once we have manufactured 500,000 units. After one million units, the price will fall to around $100." Illustrates how much R&D
they might have to pay down... and another arguable 'error' of a company warning away early-adopters, though I do find that sort of honesty (and bounty approach) more refreshing than... say, Apple.