Totally agree with you; my point is that for replaying old games, RPi is well enough. No point in making what is esentially a hardware emulator and calling it "the real thing".
Ah, I see. On that, there are different degrees of "realness". Aside from historical preservation, there are enthusiasts (I might be one of them :) ) who want the most realistic possible retro experience given the constraints of ageing hardware.
Using the actual hardware with 100% accurate modern part replacements is one way to get there. Another way is to have perfect, "cycle accurate" emulators but these aren't easy to come by - some emulators have been in development for years and still have glitches. This kind of hardware replacement has a much higher chance of reaching the 100% accuracy I'm referring to.
The problem is that FPGA's are not 100% hardware replicas of the chips. The only way to be close to tru experience is to use newly made silicon, but made according to the blueprints from 1970s. For example 6502s sold today are of this type; also 74AC/HC/LS chips.
These are for the most part digital chips. If you make a cycle accurate replacement, they're close enough in that they for most users won't change the output. E.g. here's an in-progress cycle accurate 6510/8500 replacement, that is if anything more compatible than some of the alternative "period" licensed 6502/6510/8500 designs, where support for the undocumented instructions vary:
There are other similar projects. This notion that the use of FPGA's means there's some inherent difference in fidelity that newly made silicon wouldn't have is nonsense for parts like this that are so far from pushing the limits of available FPGA's.
They are not hardware replicas (as in "the same circuits"), but as far as the pin voltages are concerned, they can behave exactly like the original chips. And when I say "exactly", I don't mean "pretty much the same", I mean the same. That's the beauty of the digital world, it's absolutely possible to fully and perfectly emulate digital circuits.
Arguably an emulator could do the same, but it's much harder to simulate the entire system than it is to isolate a single chip and emulate that.
It is not the point that replica behaves "perfectly same", the issue is that they are not period accurate. Even it is a perfectly
But also keep in mind that there is no such a thing as a "digital circuit" in physical world, because there some analog effects at work in any digital circuit. If it were not true, the would not have been hardware incompatabilities seen even in moderrn systems, which often happen due timing/delay or noise issues. WRT to retro tech, in for example Apple-I modern 74hct/act chips would not often work compared to more humble 74ls chips. There are also interesting temperature-dependent effects in NMOS 6502 (some undocumented instructions are executed differently) which have non-digital reasons and cannot be reproduced.
Emulators are a lot easier btw to write than a VHDL or verilog spec and put it to actual hardware and debug etc, due to having already so many existing open source code for emulators, processors etc.
My point is different though - there is no way to make new Commodore 64 out of existing components. What you will get is a faithful replica with zero historic importance. For some it does not matter, but I still believe that even these people make a majority, the resulting product is not C64 and should not be called such.
