Hi,

I made a 6502-based computer for high-level programming languages.
I also wrote a tiny compiler for the computer.

https://github.com/hatsugai/Fenestra6502

hatsugai

Welcome! I came across a mention of your project yesterday and was intending to post about it.

It's a highly interesting idea:


Very nice diagrams and explanation too - thanks for that and well done! (And thanks for choosing an open source license)

Several people here have a great interest in circuits which expand the capability of the 6502, and several have an interest in high level languages.

Welcome, hatsugai, and thanks for posting! (And thanks, Ed, for excerpting some highly pertinent detail, indispensable for an introductory post on the subject.)

hatsugai, I compliment you on the conception, execution and writeup of this project! It's a good idea that you've tidily implemented then carefully explained. Bravo!

(Off the top of my head, other circuits which expand the capability of the 6502 include Marc Brooks's radar controller, Acorn's in-house extended-address machine, and my KK Computer. Documentation is always a challenge, though, and I think I could learn a thing or two from hatsugai! )

cheers,
Jeff

_________________
In 1988 my 65C02 got six new registers and 44 new full-speed instructions!
https://laughtonelectronics.com/Arcana/ ... mmary.html

Thank you for your kind comments.

hatsugai

It's an interesting idea. You're basically creating a hardware data stack, which given the 6502's page zero addressing modes also acts like a register file.

It's easy to rationalize the results for all except the stock 65C02. This doesn't reflect well on the comparative abilities of cc65 to emit performant code (or is SDCC just unusually good?).

_________________
Got a kilobyte lying fallow in your 65xx's memory map? Sprinkle some VTL02C on it and see how it grows on you!

Mike B. (about me) (learning how to github)

It might be saying that it's easier to make a performant C compiler for a Z80, but I think that's a different topic for this thread - a good idea for a new one perhaps?

I fixed a number of mistakes on the schematic.

hatsugai

The benchmark result of MC6809 is corrected. (I miscopied.)

tarai: 54.421 sec -> 51.421 sec

hatsugai

Thanks for the update! You might see the discussion over here
viewtopic.php?p=78840
using fib and tarai benchmarks on various platforms.

I had to look up the tarai benchmark:
https://en.wikipedia.org/wiki/Tak_(function)

But note that tak() is subtly different from tarai(). See also this pdf.

Thank you for the information.

I was surprised to see the deep discussions.

I seem the tarai function is suitable for this kind of benchmarks, especially for measuring the performance of procedure/function calls.

Dr. Ikuo Takeuchi, the inventor of the tarai function, is famous as a Lisp hacker (one of the developper of TAO and NUE). It is said that tak version is McCarthy's mistake.

By the way, I made a MC6800 SBC (MC68B00, 2MHz) to compare the performance with Fenestra 6502 and others. Does anybody know if there is a C compiler for MC6800? If not, I will write a code generator of CCLV for MC6800.

hatsugai

Decided to post a bigger benchmark post and moved this one over to it - everything you wanted to know about benchmarking (well... maybe not everything!) is here:
viewtopic.php?f=1&t=6323

-gordon

_________________
--
Gordon Henderson.
See my Ruby 6502 and 65816 SBC projects here: https://projects.drogon.net/ruby/

I think there might not be one. (HP had one which ran on their Unix workstations and on VAX/VMS, but getting that working would be difficult.)

A new code generator could be of interest to many...

I wrote a code generator of CCLV for MC6800 and measured the execution time of fib and tarai.

Interesting! (Your CCLV repository is here.)