Demos and Tests of my Fast 6502 System

Here's a tour through some demos and test of my 6502-based system. The demos were intended to showcase its speed, but the flaw in that plan was not really being able to compare that against anything!

Some of the demo code is available on github:

00:00 Introduction
00:24 Life
01:03 Serial bootstrapping
01:30 Python-based serial server
02:03 Mandelbrot
02:54 Prime Sieve "Drag Race"
04:34 Gordon's Interactive BASIC Language
05:55 CPU Speed measurement
07:07 Memory test
08:24 Klaus Dormann's test suite
11:03 Video output plans
12:40 Second lease of Life
13:45 Video circuit interfacing schematic
14:10 Even more Life

[microhobbyist]

Impressive clock speed. I must have missed the explanation from previous videos, but how did you determine the max clock speed? Trial and error?

For video, to achieve the best possible bandwidth, I vote for no more than 640x480 in text mode, and 320x240 in graphic mode with the appropriate amount of colors in each mode.

[charlesjmouse]

Very interesting indeed... I recently discovered your excellent channel and have been binge-watching.
On the subject of output, I realise this suggestion might go against the grain: Might you consider one of the Ti family of GFX chips?
The later V9958 (Yamaha) can do most of what you might be after, admittedly not quite at VGA. They are easy to interface with existing systems for a quick solution, and available (RC2014) schematics will aid development - the more interesting capabilities might spawn some ideas of your own.

As an aside I've seen these chips interfaced with Z80 and of course Ti based systems, but I'm not aware anyone has tried with a 6502-alike so that might be fun. They do expect some fairly old school RAM, but I suspect nohting you couldn't deal with... I just used the original RAM rather than thinking.

I botched a V9958 in to a Coleco ADAM along with an AY sound chip and a 9918 in to a PCW 8512 utilising the overlay feature a while back - the former to get an 80 col mode and widen compatibility with other retro systems, the latter to add a splash of colour - no cleverness on my part, just repurposing available RC2014 schematics and (mostly CP/M) code.

[GregCoonrod]

I would be very interested to see a deep dive on the source code for some or all of the example programs.

[thecodesorcerer]

Great machine, looking forward to the video integration. Hope you will do video reads as well.

[lunabell-]

I love your videos, especially those about graphics cards, it made me want to build my own system!

This video about the 6502 is a nice demonstration of what the 6502 is capable of doing by itself!

[GodmanchesterGoblin]

Fascinating! Thank you. In the schematic at the end of the video I noted the two HC259s configured into a 4 to 16 high-active decoder with no extra glue logic. I never came across that mode of operation before with the 259, but it's a very clever way of handling the decode from the FIFOs. I look forward to seeing where this goes next.

[xotmatrix]

Very cool. Based on my experience writing a cellular automata engine for the Apple II, I reckon your Life algorithm could be 100x faster with some lateral thinking.

[john_ace]

I really liked your video, thanks!
Regarding graphics: I like 640x400 in B&W (32Kb/32000bytes VRAM) as well as 320x200 in 16 color (32Kb VRAM). It can fit nicely into the 64K boundary and can be easily mapped into memory. If you use the remaining 768 bytes of memory in both resolutions, you could add palettes with 16 bit accuracy (16 palettes of 16 colors = 512 bytes). A 4 bit color index (16 palettes) per line (200) will result in additional 100 bytes of memory. There would be 156 bytes remaining for other features although some registers (resolution selection) are needed. I like how this method uses nearly all available ram.

[GordonHenderson]

And I never thought anyone would pronounce GIBL like Jibble. I say Gibble like GIF or Gibbon.
Looks good though! Thanks for running GIBL 🙂

[stevenbliss]

I like your reasons, and in spite of my love for elegance, I likely would have done the same. :)

[Dream_scape]

This is Awesome, can you make it run Arch btw ?

[LifewithDavid]

Well done! I now have a new challenge. I was hoping to write an emulator that is as fast as 6502; you just raised the bar! Can't wait to see where you go in the future.

[stevenbliss]

For video RAM read back, use "shadow" RAM which is written to at the same time, but is the ONLY one access during read.
...I assume this is a configuration your are considering?

[stevenbliss]

Why did you not make the CPU on the "motherboard"? I know DEC mini computers, with their MultiBus architecture, also had the CPU board that way, but even though I LOVE LOEV LOEV DEC minicomputers, I always hated that part of their design, even at the age of 16 when I first came across them at school in 1976. As an electronics design engineer, I love elegant designs, and you have done that, again KUDOS to you man! :) ...but through my eyes, not having the CPU on the main board is ugly, ...hence my question???