Huh. Microchip has a PIC24 (16 bit CPU, supported by gcc, not as "nasty" as their 8bit micros) with 128K of program space and 8K RAM, in a 28pin DIP package! Makes one contemplate defection...
(Not having that DIP has been a real thorn when it comes to bigger, faster, CPUs.)
http://www.microchip.com/wwwproducts/Devices.aspx?dDocName=en534548
Makes one contemplate defection
Not this one.
I was a PIC guy back then. The only drawback they have is lack of GCC support (yeah I know their commercial C compiler is based on GCC). They have so many microcontroller to chose from!
I'm pretty agnostic towards micro-controller manufactures. As long as I can program in a higher level language then assembler and the chip is fast enough and has enough memory and hardware features and affordable, I can be 'converted over'.
I just think the Arduino folks were the first to build on all open source tools and IDE, multi-platformed, and threw in open source hardware to boot. I'm happy for their success and really enjoy their platform. However I don't think they are the only path to happiness, now and into the future. Let competition do it's thing and let the customers vote with their wallets, I say. 
Lefty
I prefer to get to know a family well and having done so it's too damn hard to change over. I know if you stay at a high level it doesn't matter but most things I do require dicking with the hardware and it's just too much of a learning curve to start over with anothr chip set.
That PIC does sound nice though 
Rob
It's not quite as nice as it could be. Apparently the "24" in PIC24 refers to the instruction width, so 128kbyte is only 43k instructions (still, a 32kbyte arduino is only 16k instructions...)
I didn't necessarily intend that this be from "someone" who wanted to compete with the official Arudino team. In theory, the team could do a PIC-based board as a followon themselves. (I don't see any other alternatives for more than 32k of program space in a DIP socket. It depends on how important you consider that feature compared to the architecture...)
The only drawback they have
Among the drawbacks are numbered...crazy banked RAM.
Among the drawbacks are numbered...crazy banked RAM.
Oops, I forgot that. And also some software based stack on some of the microcontroller.
If you're going for PIC24, at least make it one with USB on-the-go support:
http://www.microchip.com/wwwproducts/Devices.aspx?dDocName=en536121
westfw:
It's not quite as nice as it could be. Apparently the "24" in PIC24 refers to the instruction width, so 128kbyte is only 43k instructions (still, a 32kbyte arduino is only 16k instructions...)I didn't necessarily intend that this be from "someone" who wanted to compete with the official Arudino team. In theory, the team could do a PIC-based board as a followon themselves. (I don't see any other alternatives for more than 32k of program space in a DIP socket. It depends on how important you consider that feature compared to the architecture...)
Staying in the Atmel AVR's you have ATmega 164p, 324p, 644p, and 1284p, all of then are 40 pin DIPs, ranging from 16K of flash up to 128K of flash, the 644 is a very nice chip offering 4k of SRAM and its relatively cheap.
For the banked ram in PIC's, well you programm in C, so you dont need to care about that and they already have gcc compiler for almost all their ranges, so making a fully open-source compiler for pics is much easier than say at 4-5 years ago.
For the banked ram in PIC's
Essentially no longer present in the 16bit PICs (PIC24, PIC30, PIC33) (no more so than for AVRs, essentially.)
That's why they are able to use gcc on those. Trying to get gcc to produce code for the 8bit pics is nearly impossible, though.
Its 3V3 +/-10% supply only and its outputs can sink all of 4mA....