That's pretty damn awesome, but I bet he wishes he'd stuck with a 4-bit one like everyone else. Also, was it built by hand, or with a map editor?
Sticking to 4-bit logic, and the intent to make something that can run standard code... how about a virtual intel 4004? About the most basic microprocessor ever... add in the companion 4001 thru 4003 chips (I think they were RAM, ROM and clock gen?) and you could probably implement some kind of enormous desk calculator or the like, including pressure plates for the number and operator keys, and a simplistic 7seg-like output, rather than having to flip endless switches and read binary off torches. Plus the speed would be more suitable.
(However, was there ever an "actual" computer made with the 4004? Anything that could run a recognisable game, for instance? Even tic-tac-toe?)
It's always the problem with these things that it's quite easy to emulate a very simplistic system - such as building, or even prototyping something on a 4004 level with breadboards and transistors - but going anything higher than that would be backbreaking at best, more likely impossible, and stress the simulation environment's physics and render engine (or indeed, be so large IRL that signal propagation time and strength actually does become an issue) so much that it'd be heinously slow.
Make it RISC and low-bit, in the name of speed. It's not like making a 16 bit one will confer much advantage
I mean, just imagine trying to make a virtual 68000 to take advantage of it. A Z80, 8008/8080 or 6502/6800 level cpu would be hard enough to simulate (8 bit internal, 16 bit external).
(hey, why not 6-bit? there were enough 60s/70s era minicomputers built on 12, 18, etc bit width technology. there's no rule saying you have to stick to powers of 2. 64 addresses of 6 bits each gives you the equivalent of 48 bytes to play with, vs 8 for a 4-bit or 256 for 8-bit addressing, and plenty enough opcodes, display capability, etc without being as exponentially massive. Go up to 12 bit multiplexing I/O if you need to - a good 3kbyte's worth of memory space)
The metaminecraft thing... I can't see it working. You're going to have to simulate something of the level of an atari VCS at the very least, which would push the very boundaries of the 4/6 bit systems posited above, and require several days to grind through the same number of clock cycles as even that would manage in one second (it ran at a few hundred khz). For something more like what the thread OP actually envisages, you'd need a Commodore PET, Sinclair ZX81, etc... 8-bit with the equivalent of 64x48 pixel displays and at least 1kbyte RAM (preferably 4, 8, 16k), and 0.75 to 4mhz processing speed. If the example circuits are chugging through about three cycles per second on a decent machine (30fps) and according to the official literature can go up to maybe 6 (16? i forget. still not many) cycles/sec on a fast one (100+ fps), even 750khz (zx81 slow mode, as used for drawing the scene on, e.g., 3D monster maze) will take a scale speed of ~46000 seconds per game-second. Or in other words, 13 hours. If we're cramming 4mhz into 3 cycles/sec, well, that's 15 days... 3DMM needed about a half second to draw each update (and several seconds in "fast" (3mhz) to generate the game world with no display)... are you willing to wait a month for the game field to be generated each time you want to play, and a week for each move to take place? It'd be like playing chess by mail. Tedious and impossible to debug.
Ah, now there's something that minecraft could probably handle... computer chess. 16x16 board, only 6 different kinds of piece (12 when colours taken into account - can be represented with a 2x2 torch array, for a 32x32 board) and quite simple play rules. Might not be able to have any AI, but certainly checks for illegal moves, automatic piece removal, win/lose/check/stalemate conditions etc.
Or hey ... why not some seperate redstone circuit simulation utility. You could build your circuits in it and see what they could do at much higher speed... and at least say "this COULD run in minecraft, but until we have much more powerful PCs, it's not practical".
Alternatively it'd be an interesting environment to explore the limits of what massively parallel processing can do for you. It'd be a bit like the human brain. Maybe 10 ticks per second, but each tick deals with a great many different processing tasks simultaneously... and is also somewhat asynchronous. Maybe we could simulate an ant brain.
Arrgh my head
Sticking to 4-bit logic, and the intent to make something that can run standard code... how about a virtual intel 4004? About the most basic microprocessor ever... add in the companion 4001 thru 4003 chips (I think they were RAM, ROM and clock gen?) and you could probably implement some kind of enormous desk calculator or the like, including pressure plates for the number and operator keys, and a simplistic 7seg-like output, rather than having to flip endless switches and read binary off torches. Plus the speed would be more suitable.
(However, was there ever an "actual" computer made with the 4004? Anything that could run a recognisable game, for instance? Even tic-tac-toe?)
It's always the problem with these things that it's quite easy to emulate a very simplistic system - such as building, or even prototyping something on a 4004 level with breadboards and transistors - but going anything higher than that would be backbreaking at best, more likely impossible, and stress the simulation environment's physics and render engine (or indeed, be so large IRL that signal propagation time and strength actually does become an issue) so much that it'd be heinously slow.
Make it RISC and low-bit, in the name of speed. It's not like making a 16 bit one will confer much advantage
I mean, just imagine trying to make a virtual 68000 to take advantage of it. A Z80, 8008/8080 or 6502/6800 level cpu would be hard enough to simulate (8 bit internal, 16 bit external).(hey, why not 6-bit? there were enough 60s/70s era minicomputers built on 12, 18, etc bit width technology. there's no rule saying you have to stick to powers of 2. 64 addresses of 6 bits each gives you the equivalent of 48 bytes to play with, vs 8 for a 4-bit or 256 for 8-bit addressing, and plenty enough opcodes, display capability, etc without being as exponentially massive. Go up to 12 bit multiplexing I/O if you need to - a good 3kbyte's worth of memory space)
The metaminecraft thing... I can't see it working. You're going to have to simulate something of the level of an atari VCS at the very least, which would push the very boundaries of the 4/6 bit systems posited above, and require several days to grind through the same number of clock cycles as even that would manage in one second (it ran at a few hundred khz). For something more like what the thread OP actually envisages, you'd need a Commodore PET, Sinclair ZX81, etc... 8-bit with the equivalent of 64x48 pixel displays and at least 1kbyte RAM (preferably 4, 8, 16k), and 0.75 to 4mhz processing speed. If the example circuits are chugging through about three cycles per second on a decent machine (30fps) and according to the official literature can go up to maybe 6 (16? i forget. still not many) cycles/sec on a fast one (100+ fps), even 750khz (zx81 slow mode, as used for drawing the scene on, e.g., 3D monster maze) will take a scale speed of ~46000 seconds per game-second. Or in other words, 13 hours. If we're cramming 4mhz into 3 cycles/sec, well, that's 15 days... 3DMM needed about a half second to draw each update (and several seconds in "fast" (3mhz) to generate the game world with no display)... are you willing to wait a month for the game field to be generated each time you want to play, and a week for each move to take place? It'd be like playing chess by mail. Tedious and impossible to debug.
Ah, now there's something that minecraft could probably handle... computer chess. 16x16 board, only 6 different kinds of piece (12 when colours taken into account - can be represented with a 2x2 torch array, for a 32x32 board) and quite simple play rules. Might not be able to have any AI, but certainly checks for illegal moves, automatic piece removal, win/lose/check/stalemate conditions etc.
Or hey ... why not some seperate redstone circuit simulation utility. You could build your circuits in it and see what they could do at much higher speed... and at least say "this COULD run in minecraft, but until we have much more powerful PCs, it's not practical".
Alternatively it'd be an interesting environment to explore the limits of what massively parallel processing can do for you. It'd be a bit like the human brain. Maybe 10 ticks per second, but each tick deals with a great many different processing tasks simultaneously... and is also somewhat asynchronous. Maybe we could simulate an ant brain.
Arrgh my head
