3 ms·
All these machines end up involving one or more FPGAs -- to varying degrees -- simply because of the requirements and logistics. There's varying types of purism
by cmrdporcupine 6y ago
All these machines end up involving one or more FPGAs -- to varying degrees -- simply because of the requirements and logistics. There's varying types of purism involved ("FPGA only where we would have used a custom chip" or "FPGA but not for the CPU" etc) but once you start down that path the whole thing starts to seem silly as with many of these machines the entire functionality could be done on a single relatively inexpensive FPGA but isn't for purist reasons. And then you start realizing how arbitrary many of these projects are. They're admittedly nostalgia projects, but everyone's definition of what nostalgia is for them differs. Attempts to make it more "canonical" are made by locking down the FPGA use, or keep the source to it closed, but again... arbitrary.
In a way this is why I'm personally more interested in the MEGA65 project because it's all-in on the FPGA angle, all open source, and has focused more on the stuff that makes these projects actually logistically tricky: cases, keyboards, packaging, distribution, toolchains, OS, etc. And backwards compatible with the C64/C128, which is something the X16 and the C256 can't be.
It's taking a long time to finish, but I think it's really neat. And I feel like the developers are taking themselves a little less seriously/dogmatically and seem fun.
https://mega65.org/ https://mega65.org/
- II2II 6y agoIn a lot of cases, they aren't left with much of a choice. Many chips aren't in production any longer, either due to them being custom chips or due to them serving a need that the market no longer has. Stuffing everything into an FPGA isn't really a solution in this case since it effectively makes it a hands-off machine. What do I mean by hands off? In terms of software, it doesn't make much of a difference. You can still develop for the machine and it will appear to be the genuine article. In terms of hardware, it is a different ballgame. At best, slots and ports may be exposed so stuff can be added to the machine. It will be a bit like using a modern desktop, albeit with the opportunity to develop your own expansion cards since the expansion bus can be more accessible than PCI-e. On the other hand, most modifications to the hardware itself must be made through the FPGA. The FPGA route will be more flexible, but it requires an entirely different skill set. Relatively few people have the skill or interest to deal with FPGAs, particularly if they are seeking something more comforting than modern hardware. How is replicating something like the video chip with an FPGA somehow acceptable? Presumably it is because it has to interface with the rest of the system in a conventional way and since people were unable to modify the video chip in vintage computers anyhow. Since it is interfacing with the system in a conventional way, you can also mess around with the video circuitry surrounding the chip.
- wvenable 6y agoI'm building a hobby computer from scratch with a modern 80s era CPU. Anyone doing this kind of thing as a hobby has to decide what kind of project it is, what you want to get out of it, and where you draw the line between modern and old. For me, I had never done electronics before so actually wiring up the components and doing logic at a electrical level was interesting in itself. But also I'm interested applying my programming knowledge to the limitations of that era. It's sort of an art project. No matter what, a computer today with an 8Bit CPU isn't practical. I'm going to have a 6502 CPU, I'm going to try for 1MB of RAM banked, SD card for storage, and I'm thinking of using a Pi Pico as a VGA graphics card. Obviously using a Pi Pico in a project with an 8bit CPU is "silly" but that still satisfies my own goal. Other people working on these types of computers tracking down chips from that era or are building their own TTL-logic basic graphics cards because they have different goals.
- hakfoo 6y agoWhy not a 65816? That gives you the 6502 compatibility and a larger address space mode when you want it. In a way, an anemic CPU makes sense if you go back to the old Commodore model of smart peripherlas. You don't need a lot of CPU horsepower, for example, if you can communicate with the mass storage/network card/etc at a high level and it just DMAs into place.
- cmrdporcupine 6y agoThe bus de-multiplexing with the 65816 is a pain for amateur hobby stuff. A W65C265S is worth looking at, tho it has some annoying masked ROM shipped on it you have to work around. It's basically an 65816 + UARTs + PIA + other stuff all packaged together. It has a boatload of pins which can be programmed to be either GPIO or the full address bus of the 65816 (instead of having demultiplex them like on the real 816). Comes in PLCC format, which is relatively hobbyist friendly as you can get adapters for breadboard, and it can be done thru-hole. Only downsides I found are the onboard ROM (which you can shortcircuit around I believe) and its maximum bus speed is 8mhz, lower than the 14mhz of the 65816.
- erosenbe0 6y agoPretty cool. I am not too knowledgeable about fpgas but the 144 qfp or bga parts have so many pins. If you have 32 clocks per clock of emulation that is plenty leeway to have an awesome debug header and monitor that was never possible with the real thing.