nesdev.org

Now that the 8bitguy is making his own machine with a 65816 in it, the number of people pledging to make a C compiler for it seems to have gone up. If it actually happens or not to be seen, but we might at least get more 65816 tools out of it.

Please link to those. I have no idea who is 8bitguy, and googling for 8bitguy 65816 shows some dream computer blog post from last Oct.

He's a famous YouTuber in the retro gaming circle, and he's building that dream computer now.

https://www.youtube.com/watch?v=ayh0qebfD2g

The video shows a bit of his process writing a c64-like kernel for the 65816 and modern peripherals and is outlining general features.

Design goals are:
-low price
-direct hardware intefacing just like with the c64
-modern storage medium
-no NOS components, just readily available off the shelf parts
-no FPGAs

One of his to-solve problems is what sort of GPU/PPU to use. Honestly i'm not sure how you'd solve having a video chip like the ideals he describes without using an FPGA. The current solution is serial, but he wants something that's parallel. The dormant "open PPU" project is not too far of from what he wants, but that'd break the no programmable chip preference.

I'm sure he can solve audio with some modern synth or sample stream chip, but i'm a little worried about the or convoluted interfacing with more modern synth chips. Choosing a practical-to-write-music-for soundchip is going to be crucial to people actually using it.

Edit: fixed autocorrect.


Edit: I see the prototype discussion group on FB has 5000 members and 90+ new posts just today. He's enthused quite a lot of people with this project.

Yeah the FB group is a ummm mess.. is the best way I can describe it. He didn't get what he was really hoping for But there are more people looking at the machine than say the Mega65, Phoneix256 and SpectrumNext.. which means it does have a better chance of going somewhere.

He has since relented on his no FPGA for the Graphics chip, as it is impossible to not use one.

Oziphantom wrote:He has since relented on his no FPGA for the Graphics chip, as it is impossible to not use one.

So there goes the "inexpensive" requirement as well, right...?

The gamedrinou https://excamera.com/sphinx/gameduino/ which is what he is thinking of using is only $39, and that is own board etc, once you put the parts on to a custom board with the rest of the machine, it will probably be cheaper. However it uses an Spartan which is obsolete, and people are looking at porting it to one of the new cheaper FPGAs you can get. Since I think Bil Herd is supporting he will probably push them to Altera as he hates Xillnix with a passion for that one time they screwed over Greg Berlin on the A4000.

He didn't get what he was really hoping for

Yeah it's a double edged sword. On one hand, you'd get better signal to noise ratio if you actually turned to 65xx developer forums (like this and others)... a lot of the bicycle shed syndrome going on in there from what i saw is also because people think it's a bicycle shed, when it's really not.

On the other hand... if you only turned to current 65xx developers, you have a limited circle of potential users - easily leading to abandonment. While current 65xx users might be interested in using it, the way i interpreted his video is that it'd be nice to have a modern computer (no tape loading horrors etc) that really teaches you programming naturally and is fun to use because its only layer of abstraction is a focused kernel. So i think it's safe to say that a big portion of the user base is meant to be people not already programming assembly, or not already programming much at all but are interested in starting out.


One thing i've thought about today is... i really hope this goes well. But for it to reach its potential, it would probably help to have some "killer apps" beside a BASIC interpreter at launch, perhaps sold as addons or bundled in or distributed for free, whichever model works for each contributor. I think you need a software ecosystem to start with, in order to keep it living. Some good games, in short. For it to have that, some current 65xx developers and teams need to get onboard. For current devs to get onboard, porting already existing independent games from the NES and c64 scene need to be easy enough. Then, i think you want the graphics chip to be able to do something quite similar to what the NES PPU or VIC-II does. Partial backwards compatibility or parallelism would really help.

Someone is bound to port DOOM (someone always does for every system), whether it plays nice or not. But what about after that?

The Parallax Propeller 2 is due out this year, for real and not a lie.
I have an engineering sample sitting next to me.

It's 8 cores, 32-bit wide instructions, 200+ MHz clock speed.
Two clocks per instruction when executing from the 512 long local core memory or when its streamer cache is full executing from shared central memory.
Has bit-level instructions that use a barrel shifter, hardware multiply instructions, pipelined CORDIC (trig functions), USB capability, SD card boot capability.
8-bit DACs on every one of the 64 I/O pins with ability to dither to effectively about 12-bit, so VGA or component video is easy.
Audio would effectively be 16-bit per channel after filtering.
DVI signaling over for example an HDMI connector is possible.

So what does this mean?
At the very least, it could be applied as a less expensive peripheral interface, mass storage interface, graphics chip, and sound card instead of an FPGA.
Modes like palleted 1024x768 VGA won't break the memory bank and require an outside DRAM chip.
There is already a SID emulation (SIDCog) available on Propeller 1, recently ported and working even more faithfully on P2.

As for its long and battered development history, just yeah.
Oh so very yeah.
I'd say the original Propeller 2 died of "second system syndrome", and this refocused version is really #3 as it quite honestly started over from scratch.
But marketing is tricky.

1024x768 = 786432 and we have 1 byte per pixel so that is 786432/1024 = 768K
4bits per pixel = 384K
2bits per pixel = 192K
1bit per pixel = 96K

So if it gets 512K then you can have 4bpp, gives a 128K but you still need to store sprites as well. Ok for a "title screen" but not really a game.
256K 2bit is possible but again not leaving much for sprites
1bit is possible for 256K though.

I would think its probably going to 640x480 max, which is 300K at 8bits and 150K at 4bits. So I would think 640x480 16 colour is going to be the practical limit, and maybe it will need 320x240.

It would probably be better to change the thread from "65816 C compilers" since the discussion's gone elsewhere.
Anyway, my hot take is that I don't really see a target developer audience for this. If you're willing to put up with an antiquated 33 year old processor, you probably have an existing platform in mind. If anyone here disagrees with me and wants to write software for a new 65816 based computer, I'd love to know why.

I believe that old computers/consoles still have a following today because of nostalgia and/or sizeable software libraries. A new architecture will have neither, so I don't see this having much support once the novelty wears off.

FWIW, I share the same sentiment. I think this is a personal project which, at best, will become similar to CHIP-8. Obviously making improved software that runs native to the system/CPU (presumably 65816) nobody will take issue with, but I strongly doubt this will somehow pull the 65816 from the ashes and make people start developing tons of new and amazing software for it; I expect a short-lived "trend". Honestly, the only reason I can see here that the 65816 was chosen was because the inventor is already familiar with 65xx architecture -- and that's fine (hopefully obvious to readers?). I am certainly not going go through Facebook to try and "catch up" on the decisions or history -- it's not my project, so my opinion/view really has no bearing.

This project really just has me thinking about the possibility of developing a sort of old "8-bit" computer style kernel for an SNES flash cart... Not having a keyboard is a bit of a problem though...

Drew Sebastino wrote:Not having a keyboard is a bit of a problem though

Cellphones solved that problem... and/or you could put buttons on the cartridge!