8-bit Guy is designing a piece of hardware

I figure here would be interested not just in the project existing, but might have the knowledgebase to assist in some fashion.

I think I hear that this project is proceeding in a closed Facebook group, so I’m out of the loop there. But I think I gather that it will be an '816 machine, with some effort perhaps to be C64 compatible? And with a graphics engine of some kind - in FPGA?

I do know he doesn’t want to go the FPGA route for anything, but at the same time I don’t know what reality will impose on his ‘ideal situation.’

His Facebook page is active, full of “will it do this”, “SID?”, “why not use this graphics chip” and so on, however from what I can gather, he more or less fixed the specification 6 months back and that’s been published. He is using an FPGA though - as part of a Gameduino for the graphics side of it all and yes, it’s 65C816 based.

It’s also worthwhile mentioning two other public 65C816 projects too - one is called the Foenix and there are some videos of its creator, Stafanie, with 8-bit Dave very shortly after Dave talked about his “ideal 8-bit computer” on his youtube channel - that project is sort of closed, but there is a website with not a lot going on, however I understand there are now prototypes for the early “core” developers - but again, it’s yet another system inspired by the C64, however the designer has gone quite overboard with a pair of FPGAs, stereo sound, DMA engine, joystick ports and some othre stuff.

The other public (ish - closed facebook group) project that I’m aware of is the Neon816 - again, a 65C816, some RAM, big FPGA, joysticks, 3.5" floppy and so on. The creator of that; Lenore, seems to be a Forth enthusiast, but who really knows.

-Gordon

2 Likes

Ah Neon816 does have some public activity:
https://hackaday.io/project/164325/logs

I posted about the Neon 816 here:

It has limited info available, but does look very interesting.

1 Like

Oops, so you did. Who would have thought we needed to search for dupes before posting, so soon! (It’s not even the first dupe…)

In fairness to the chip selection? the 65C816 is Mature, Backward compatible with it’s older 8 bit kin, and still being produced. That last is vital since ‘harvesting from old machines’ is inherently unreliable and teaching people about the older hardware and programming techniques they fostered you are going to compete with things like the raspberry pi and other single board computers that are insanely cheap so ‘oh hey we can get this thing for a couple hundred bucks’ is going to be seen largely as a tinkertoy for people with too much money.

Hence why I hate the idea of using SID for sound. Yes its iconic yes it’s a good chip, but it isn’t being produced.

I met personally with David Murray (The 8-Bit Guy) for a few hours on Sunday. He is just a figurehead for this project, but isn’t too deeply involved with it.

They are using a 65C816S, 6522, an FPGA with a Gameduino core converted to be accessed on the parallel bus, not by SPI, 1MB of RAM and at least one SD slot. It will support 16 and 256 colors and a good sprite capability. The system voltage will be 3.3v. It will have a stacking-type demuxed expansion bus.

I have committed to design a 2M SRAM and 12M DRAM expansion for it.

2 Likes

I thought he was writing code?

Honestly, from what he said, I got the impression he would port his games to it, and write something new for it, but the OS or the kernel or language it uses, that has all been delegated.

He’s a figurehead enough that he wasn’t really up do date or aware of what the team had been doing and saying the previous several days.

I’d say he was more “specifications and documentation”…

1 Like

When I was following this in the beginning, many many months ago, he was talking about writing kernel stuff.

How did you even get to meet with him?

Bo Zimmerman lives just around the corner from me. He loaned David the Commodore PC-10 for his video a few weeks ago. He drove down to visit friends in Austin and collect it, and we hung out and chatted for about four hours.

Met his missus too.

2 Likes

I just read the latest update from David, and it sure as heck sounds like he’s kernel hacking.

1 Like

It sounds more like they’ve had additional issues hardware hacking by the looks of it - ie. sticking to a 65C02 with 64KB of RAM (although it will have more RAM, paged in in 8K chunks), rather than using the 65C816 in 24-bit address bus mode. (So it will be possible to plug in an '816, but only use it in 16-bit address-bus mode).

The '816 does have some timing quirks that may make it challenging to use those extra 8 address lines, however there is a standard published circuit by WDC to do it. There is technically a timing issue as you have to latch the output of the CPUs 8-bit data bus to form the top 8 bits of the address bus - but at the exact same time, RAM could be responding to a read cycle and thus writing its data to the data bus, thus corrupting the 8-bits which are in the process of being latched to form the upper 8-bits of address bus.

It’s complex, but Apple managed it in the early 80’s in their //gs and there are many other 816 projects out there (Super NES to name a commercial one) that seem to work just fine.

I have exchanged an email with Dave though - their 8K for the BASIC and 8K for the OS (Kernal) may initially seems limiting (especially when I assembled MS Basic 2.0, aka Commodore Basic 2.0 to run on my system and it came in at a shade over 9K of code, however since they’re re-writing everything then it can be optimised and use 65C02 instructions and put some of the BASIC code in the OS ROM too.

There is also 8 x 8KB banks of ROM (and BASIC is in one of them) and 64 banks of 8KB of RAM in the 64KB memory map.

The FPGA graphics chip the’re using appears to be directly bus-mapped to the system too rather than going via e.g. a 6522 style VIA.

So who knows.

-Gordon

1 Like

Note that they are going with a 65c02 for their first prototype, but after the hardware works, they claim they will try to see whether they can make a 65c816 work in it inside the 64K address space (that is, ignore the bank data on the data bus in the first part of the clock cycle, to use it like a 65c802).

It’s the version of the development path where the 65816 works out where I would be a definite backer when they go to kickstarter. For the 65c02 version, I’m on the fence.

Making a 65c186 work in a 65c02 socket is a matter of careful design then it will simply drop-in. (One pin needs pulling high on the '816, but it would do not harm to pull it high via e.g. a 10K resistor, on the '02 as it’s a clock output pin who’s use has been deprecated for some decades now)

There are a few pins on the '02 that you really don’t need to use, and as long as you don’t need the functionality that the 816 gives you on those pins, then it should be a drop-in part.

I’m going to give it a go on my Ruby 6502 board soon, as a precursor to getting my existing OS running on my 816 board which I’m in the process or laying out now.

The one thing from a software point of view is an extra hardware vector in the 816 that clashes with a software vector in my OS (which is vaguely Acorn MOS compatible). However this is the abort input - which is the pin above I mentioned that needs typing high - unless you create a design that actually uses it.

-Gordon

And almost on-cue, an update shows a render of the PCB on his proposed system. All through hole, but I didn’t think it’s expected to be built by the user…

Nothing too special, really. Just another 65C02 system with a couple of VIAs and some banked memory.

Wonder what memory he’s using as the best I could get in through-hole runs at 55ns…

-Gordon

If the program flies, it’s primary appeal is scale. That’s another reason I hope the 65816 drops in fine, since from the response within the FB group to the option between Yet Another 65c02 board and a 65c816 board, the 65c816 led the poll 2:1. Since a 65816 can be used as a generic 6502 system by staying in 6502 opcode emulation mode, it seems like it is a scale multiplier.

Their mooted project path is the through hole version for development boards and then refined for kits, then a surface mount version for the main completed board purchase, then a version with much of the side parts replaced by FPGA for the lowest cost version. So if the first board didn’t look like Yet Another 65C02 board, that would have been a substantial departure from the mooted path.

1 Like

I voted for the 6502 option there - however they didn’t give me enough information and from what I now know, they had already made the decision to go with the 6502 before they posted the poll, else, how could they go with the 6502 when their poll showed overwhelmingly that the followers wanted the 65816 option.

This, along with the posts )of mine) they removed has made me lose confidence in them, but I’m keeping some interest, so who knows.

-Gordon