On Sunday, August 27, 2023 at 10:56:10 PM UTC-7, minforth wrote:
Zbig schrieb am Sonntag, 27. August 2023 um 10:22:17 UTC+2:
30-years old idea — still very interesting, especially
in the era of fast CPUs/microcontrollers: http://wilsonminesco.com/0-overhead_Forth_interrupts/
Thanks for the link.
But FWIW I'd never blow up NEXT and polling is rarely a good idea particularly "in the era of fast CPUs/microcontrollers". Bad carbon footprint. ;-)
Polling is needed on the 65x because it has only one IRQ.
Garth Wilson continues to focus on the 65x in this "era of fast CPUs/microcontrollers." If he likes it, then good for him!
He has an indirect-threaded Forth. That is very slow on the 65x!
My 65c02 cross-compiler was based on ISYS Forth and it generated
machine-code. I wrote this for my symbolic math program that did
derivatives (I didn't do integrals because they are more difficult and
because I was at the limit of what a 65c02 can do speed-wise).
There is no Forth or C or Pascal or anything else that could have
gotten as far as doing derivatives.
Machine-code on the 65c02 is bloaty (I was using the full 128K
of the Apple-IIc for my program). To have less memory usage,
such as in a micro-controller, I would like to use byte-code.
The 65c02 was a huge disappointment! They used the X register
for their byte-code table jump. The X is used for the data-stack
pointer though. They should have used Y instead. This was the
most egregious design mistake in the 65c02, but there were other
mistakes. A more thought-out design for the 65c02 would have
kept the 65x line going --- instead, the i8032 took the microcontroller
market and the 65c02 became obsolete.
--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)