Maybe the solution is one is connected normally, and the other just get its input driven.
Paragraph 8.4 seems to suggest that would be viable:
"Full Swing Crystal Oscillator
Pins XTAL1 and XTAL2 are input and output, respectively, of an inverting amplifier which can be
configured for use as an On-chip Oscillator, as shown in Figure 8-2 on page 29. Either a quartz
crystal or a ceramic resonator may be used.
This Crystal Oscillator is a full swing oscillator, with rail-to-rail swing on the XTAL2 output. This is
useful for driving other clock inputs and in noisy environments. The current consumption is
higher than the ”Low Power Crystal Oscillator” on page 28. Note that the Full Swing Crystal
Oscillator will only operate for VCC = 2.7 - 5.5 volts."
Maybe a dual 1284 would be better - 2 user replaceable DIPs, "dual thread" action, 4 UARTs, 256K total memory, 64 IO.
Cost for 2 /1284's ($8.13) is less than a non-user replaceable 2560. ($17.97, both digikey & mouser pricing, newark $7.38/16.27).
Yep you could easily make a shield to give "dual core" functionality, one problem IMO is that there's no really good mechanism for the two to talk. SPI is close but the crap implementation on AVRs limits that as well I think.
Anyone for dual-port RAM?
Still, as long as the tasks were not tightly coupled that wouldn't matter.
How's this one look? http://www.cypress.com/?docID=24398
Somewhat affordable, $7.45 each.
Need to set up a preloadable, autoadvancing, 10 bit address counter on each side to allow fast burst of parallel access reads & writes.
Or tie up 23 pins on each side.
Yeah Cypress make some really nice (and expensive) gear in this area, I'm always loved this sort of setup and used to work with a Z8000/bit-slice processor dual board, but the truth is I can see absolutely no reason to do this sort of thing these days at most levels when you can buy a 50MHz LPC 32-bit processor for $1.50. Lash out and spend $2.50 and you get 100MHz and every peripheral known to modern man
Where's the design integration challenge in that? Now you're stuck learning some new programming language and dealing with terminating signals at RF frequencies. Might as well buy a netbook or ipad.
I admit there's no design integration challenge, but you still use C/C++ and the high speed stuff is all internal.
I've been studying the LPCs for a project, a huge amount of grunt for a chip that's just as easy (even easier in some respects) than the typical AVR we talk about here. Frankly as far as I can see it's just the Arduino/AVRfreaks communities that make it worth using 8-bit AVRs at all.
The AVRs make it all about physical interfacing. Writing some code is fine, but I don't want to get bogged down into a huge software project. Its bad enoough maintaining a website!
I didn't lay it out, was reading in the old thread about how the original developers gave up on the concept after the Mega's came out.
Was thinking I'd replace the '328s with '644s. Altho looking at the parts & connecters there, not sure two 40 pins DIPs would fit.
Looks like the '328s would fit on a 80x100mm board, looks like I could replace the resistors with smaller parts, and changing the IO pins to dual-row parts instead of 2 single row parts will tighten things up too.
You serious about making this? I normally design with 10 mil lines so I can snake 2 traces between holes if need and use 0.012" holes for vias. That gonna be a problem for home etching?
Not sure what the OP's needs are, but if they need multicore, the Parallax prop has some really nice multi-core ("multi-cog" in Propeller parlance) solutions available. If you're looking for an AVR-based challenge, carry on
Would be interesting to hear what your needs are for this - there are probably better solutions available. If your problem requires any kind of serious synchronization, shared memory, etc, FPGAs are probably a better bet (and will blow away anything you can do with the AVR, performance-wise...)