I like the version of this idea where they have a 200 MHz microcontroller on a little circuit board running a 65816 emulator much faster than a real 65816 can run, and using the microcontroller's GPIO pins to work the lines attached to the 65816 socket pins.
It evokes the image of the wasp plugged into a cockroach's nervous system, driving it helplessly to the wasp's nest. But in a good way.
Can you actually do that with GPIO pins, or would that be too high latency/jitter? Even if so, it seems like a job that would be done much more efficiently by an FPGA, though I'd guess the reason why nobody does is because there's probably other stuff tied to the clock that can't run that fast. Or does this already exist? The concept sounds cool.
It popped up on HN pretty recently. A 1 MHz bus is slow enough to operate entirely from the microcontroller's interrupt handler. The native CPU clock pin would be wired directly to the interrupt input to keep it synchronized.
One wonderful feature was that it all fit on a circuit board the same size as the original CPU package.
As I recall, it had a mode where it would copy all the slow main board RAM contents into the microcontroller RAM, and run extra fast from there, using the bus only for I/O. Seems like it must have a way to designate certain parts of the program to run at cycle-accurate native speed, e.g. to operate the Apple ][ floppy controller.
You could of course do it all with an FPGA, but that would seem more like work.
they are out of stock on W65C816S6TQG-14 chips. WDC is a very small player and they probably do die manufacturing runs every couple of years and then package according to the demand. Surprisingly Mouser has 24 chips in stock with more "promised" very soon https://eu.mouser.com/ProductDetail/Western-Design-Center-WD...
Absolutely beautiful. I love to see these FPGA accelerator cards being used as drop-in replacements for retro systems.
This makes me wonder if it's possible to make these into PCIe cards for a normal PC that would provide hardware-assisted emulation. It would be especially nice for emulating more advanced machines like PowerPC Macs.
Nostalgia, longing for authenticity. Realization a simple mouse port converter adapting USB mice to work in Amiga/Atari ST has more computing power than the computer it is used with is painful to retro enthusiasts.
My first computer used a Motorola 6809 processor, one day I opened a fire panel at work and there were sixteen of those monitoring the sensors and making the lights blink. Looked like they were driving everything with the cpus directly, didn’t see much else besides power, ram/rom, and screw terminals. Sad end for a great little processor.
Thats sorta my question too, as evidenced by the graphics, its barely faster than a transwarpGS (which is 30 year old hardware). Sure its a lot cheaper at this point because it seems transwarps are in high demand.
OTOH, I'm pretty sure you can get an FPGA to emulate a 65816 significantly faster than 16Mhz, not including the fact that making it OoO/etc might drive the IPC way up.
Speaking of upgrading to an Apple IIgs, I remember in high school in 1987 my school wanted to upgrade a classroom that had a few IIe machines to a bunch of IIgs. I begged and pleaded with them not to do it, to go with Mac's or even better PCs, as the IIgs was obvious to me even at the time as a dead end and a room full of them was a huge waste of money. They just accused me of being 'radical and embittered against Apple' because I had an Amiga 2000 at the time that I evangelized for quite a bit. I didn't quite understand their argument, as I had a II+ at home that I still used and enjoyed, and was pushing for Mac's as a viable alternative to the IIgs.
Anyway, they ran a big capital campaign and got the parents to donate a ton of money to buy all these machines. Within a year, they were collecting dust and the school bought a bunch of Mac Plus's and a few PS/2's which got used extensively.
That incident was my first lesson on the propensity of people to obstinately make stupid decisions, and so I always associate the IIgs with that concept. But so that people now don't think I am 'embittered' against the IIgs, I do actually own two of them currently :)
I mean, even the IIGS wasn't really an Apple II in a real sense. Classic Apple II programs did use the IIGS' CPU, but were otherwise basically running on the MEGA-II (which I think is also how the Mac LC did Apple compatibility, although I've forgotten). Writing programs for GSOS had more in common with Mac programming than the Apple II, and properly using all of the GS's advanced sound and graphics required entirely different code from an Apple IIe. The 65816 also had a lot of practical issues stemming from the global nature of the 8/16-bit toggle that would've only gotten compounded if you tried to naively extend the architecture to 32-bits.
In the end, yes, I'm very curious what extending the Apple II line would've looked like, in an academic sense. But I suspect that, by the mid-90s, the "Apple II" would've been as technically unrecognizable from the Apple IIe as late 90s PCs were from the original XT: lots of echoes and lots of backwards compatibility, but extremely different when properly and modernly used.
https://en.wikipedia.org/wiki/SuperCPU
https://www.youtube.com/watch?v=x9clez2fxxw