Not for the 6502, but possibly interesting to the other Forthwrights here: Foud a link on Reddit (where there is a forum for Forth) to a "Forth on a DIP" project (http://jeelabs.org/2015/07/22/forth-on-a-dip/). This uses the LPC1114 chip, a 28-pin DIP ARM (http://www.nxp.com/products/microcontro ... 4FN28.html), with Mecrisp Forth (http://mecrisp.sourceforge.net/). The author has a moment of insight: Well, duh.

Edit: Oops, I did misunderstand - there's no repeated function call, so this is a counted but empty loop taking 4 ticks for each time round. Still not much by way of overhead.

[Previously: Am I missing something? That seems quite impressive, for an interpreted language, to call a function in just 4 ticks - even if the function does nothing, doesn't it still have to do the nothing? In effect this Forth is running very close to native speed, which is about as good as it gets.]

Looks like Mecrisp compiles directly to native code (machine code), so yes, that is quite possible.

_________________
Because there are never enough Forth implementations: http://www.camelforth.com

On a related note, there is FlashForth which will run on an 18-pin DIP (Microchip PIC18) among a number of other families.

http://www.flashforth.com/

P.S. Also a short article on FlashForth in the May, 2015 issue of Circuit Cellar.

The Parallax Propeller chip has at least three Forth variants on it. The two most popular are PropForth and TachyonForth. PropForth is closer to standard Forths while Tachyon is much faster. I've done two projects in PropForth

Drawbot: https://www.youtube.com/watch?v=1Z5kE-3XzS0

ScaraArm Mk I: https://www.youtube.com/watch?v=14BX50aHDAk

There's also AmForth for several other microcontrollers: http://amforth.sourceforge.net/