26 ms·

Those machines are not comparable, the VIC uses the same memory as the 6510; while the NES uses separate memory and another CPU! (Fun fact: Commodore was suppos

by bullen 5y ago

Those machines are not comparable, the VIC uses the same memory as the 6510; while the NES uses separate memory and another CPU! (Fun fact: Commodore was supposed to build the NES for Nintendo but Tramiel bailed in the last minute, a great testament to Chucks 6502) Of course you can build a modern GPU and soundcard with a 6502, but the point here is that the VIC and the 6510 are at 1MHz and the DRAM is at 2MHz so they access it interleaved. The VIC2 was the first home GPU with hardware sprites. Also the memory access is 16-bit so it's complicated, but the audio is 8-bit and the video output is generated with 8-bit precision = very limited color = why we get the 2 pixel wide 4 color mode. So the C64 cheats; in one way it's actually not a pure 8-bit system, but the others cheat even more. The point is "the architecture of the system is as complicated/simple as it needs to be for maximum 8-bit price value and repairability". I guess that is what I meant. 64K DRAM is the most effective cheat saturating 16-bit while everyone else was struggling with worse workarounds, the NES has a pretty bad solution; shipping memory chips in every cartridge (what a waste of hardware for only anti-piracy/anti-development "gain")! Just look at the NES dev. community today, even if the NES outsold the C64 by 3x it's almost non-existent! As for sound the SID is still unsuperceeded, it's the pinnacle of 8-bit sound. Lastly the fact that you could copy software is really a big point on the + side. Piracy is the feature! Nintendos legacy is non-piracy only, even today with the patching of the bootloader bug in the Switch! Fortunately I have an original Switch and can run linux on mine! Xo

kragen 5y ago
> Those machines are not comparable No, they certainly aren't. Nor is the C64 comparable to the Heathkit H-89 I grew up on, or to the 16-MIPS AVR in my Arduino; but all three of them, as well as the SK-1, the NES, and the S1 MP3, are 8-bit computers. (The Tek 4014 wasn't even a computer, but it was 8-bit digital hardware.) Your claim was that the C64 "pushed 8-bit as far as it could go," and the admitted fact that other (existing, non-hypothetical) 8-bit computers are "not comparable" makes nonsense of that claim. > The VIC2 was the first home GPU with hardware sprites. The Atari 2600 had hardware sprites and came out in 01977, five years before the Commodore 64 and its VIC-II in 01982. (They were amazingly shitty hardware sprites, but they were hardware sprites.) Generally doing things in hardware makes them faster but less flexible, so normally you only want to do it if the CPU is too slow. An AVR or SX chip is 8-bit, but fast enough that you don't really need much hardware; I've done baseband monochrome NTSC video output in software on an 8-bit AVR with the video output hardware being only a couple of resistors. The AVR isn't fast enough to bitbang color, though. > Also the memory access is 16-bit so it's complicated, but the audio is 8-bit and the video output is generated with 8-bit precision = very limited color = why we get the 2 pixel wide 4 color mode. When we say that a computer is "an 8-bit computer", we mean that its instructions manipulate primarily or exclusively 8-bit quantities, for example because all of its architectural registers are 8 bits wide. The Commodore 64's data bus to memory was not 16-bit; it, too, was an 8-bit bus. Like every 8-bit computer I've ever seen or heard of, the address bus to memory was wider than 8 bits; it was 16 bits wide. (Most 8-bit computers had a few 16-bit instructions, too, for things like pointer arithmetic, but not the 6502.) Most digital video output is generated with 8-bit precision, or even less. You know 24-bit color, like in JPEG, PNG, most TIFFs, and most desktop and laptop GPUs? That's 8 bits of red, 8 bits of green, and 8 bits of blue. That's 8-bit precision. The reason the C64 had very limited color was that it only had 64 KiB of RAM, not because "the video output is generated with 8-bit precision". > As for sound, the SID is still unsuperseded; it's the pinnacle of 8-bit sound. The SID isn't a digital audio chip at all. It doesn't output PCM; it outputs analog audio. So it's meaningless to say that it's "8-bit". The only 8-bit thing about it is its data bus, which is 8 bits wide because that's how wide the CPU's data bus was; by contrast, its I/O registers are a mix of 1 bit wide, 4 bits wide, 8 bits wide, and 16 bits wide. (Because the data bus is 8 bits wide, you read or write them 8 bits at a time.) Its internal registers are of many different widths; the phase accumulator is 24 bits wide, for example, and the white-noise generator is 23 bits wide. Every sound card on an IBM PC ISA bus also used an 8-bit data bus, including things like the Gravis UltraSound and the Sound Blaster AWE32. Current USB and single-lane PCIe sound devices, not to mention SPI and I2C ADCs and DACs, use a 1-bit data bus. So the width of the data bus is of absolutely no consequence for the capabilities of the audio. If we're talking about 8-bit audio rather than 8-bit computers, the SK-1 (from 01985, two years later) shows you what you can do with 8-bit-wide digital audio. (It was also an 8-bit computer, but the reason its pianos, trumpets, and human voices sound like pianos, trumpets, and human voices is that they're sampled.) Or, say, Sprint's digital long-distance telephone service, launched in 01980 as United Telecommunications and turning a profit for the first time in 01989: entirely 8-bit mulaw audio. (That's not why telephone-quality voice sounds bad, though; that happens because of the sample rate, not the bit depth.) > Lastly the fact that you could copy software is really a big point on the + side. Piracy is the feature! Well, of course, but that's a feature of computers in general; it has nothing to do with 8-bitness. There's a continuing effort by bad people to prevent computers from being programmable, but it's mostly unsuccessful.
- bullen 5y ago
  Thx for clarifying things! You seem to like confusing dates, what is wrong with ISO 8601?
  - kragen 5y ago
    Sure! ISO 8601 is not Y10K-compliant. :)
    - bullen 5y ago
      Why not? YYYYY-MM-DD works!
- kragen 5y ago
  > Every sound card on an IBM PC ISA bus also used an 8-bit data bus, including things like the Gravis UltraSound and the Sound Blaster AWE32. Erratum: the ISA bus had 16 data lines from the time of the IBM PC AT, and I believe that both the GUS and the AWE32 were 16-bit ISA cards, not 8-bit ISA cards.

by bullen 5y ago

kragen 5y ago

> Those machines are not comparable No, they certainly aren't. Nor is the C64 comparable to the Heathkit H-89 I grew up on, or to the 16-MIPS AVR in my Arduino; but all three of them, as well as the SK-1, the NES, and the S1 MP3, are 8-bit computers. (The Tek 4014 wasn't even a computer, but it was 8-bit digital hardware.) Your claim was that the C64 "pushed 8-bit as far as it could go," and the admitted fact that other (existing, non-hypothetical) 8-bit computers are "not comparable" makes nonsense of that claim. > The VIC2 was the first home GPU with hardware sprites. The Atari 2600 had hardware sprites and came out in 01977, five years before the Commodore 64 and its VIC-II in 01982. (They were amazingly shitty hardware sprites, but they were hardware sprites.) Generally doing things in hardware makes them faster but less flexible, so normally you only want to do it if the CPU is too slow. An AVR or SX chip is 8-bit, but fast enough that you don't really need much hardware; I've done baseband monochrome NTSC video output in software on an 8-bit AVR with the video output hardware being only a couple of resistors. The AVR isn't fast enough to bitbang color, though. > Also the memory access is 16-bit so it's complicated, but the audio is 8-bit and the video output is generated with 8-bit precision = very limited color = why we get the 2 pixel wide 4 color mode. When we say that a computer is "an 8-bit computer", we mean that its instructions manipulate primarily or exclusively 8-bit quantities, for example because all of its architectural registers are 8 bits wide. The Commodore 64's data bus to memory was not 16-bit; it, too, was an 8-bit bus. Like every 8-bit computer I've ever seen or heard of, the address bus to memory was wider than 8 bits; it was 16 bits wide. (Most 8-bit computers had a few 16-bit instructions, too, for things like pointer arithmetic, but not the 6502.) Most digital video output is generated with 8-bit precision, or even less. You know 24-bit color, like in JPEG, PNG, most TIFFs, and most desktop and laptop GPUs? That's 8 bits of red, 8 bits of green, and 8 bits of blue. That's 8-bit precision. The reason the C64 had very limited color was that it only had 64 KiB of RAM, not because "the video output is generated with 8-bit precision". > As for sound, the SID is still unsuperseded; it's the pinnacle of 8-bit sound. The SID isn't a digital audio chip at all. It doesn't output PCM; it outputs analog audio. So it's meaningless to say that it's "8-bit". The only 8-bit thing about it is its data bus, which is 8 bits wide because that's how wide the CPU's data bus was; by contrast, its I/O registers are a mix of 1 bit wide, 4 bits wide, 8 bits wide, and 16 bits wide. (Because the data bus is 8 bits wide, you read or write them 8 bits at a time.) Its internal registers are of many different widths; the phase accumulator is 24 bits wide, for example, and the white-noise generator is 23 bits wide. Every sound card on an IBM PC ISA bus also used an 8-bit data bus, including things like the Gravis UltraSound and the Sound Blaster AWE32. Current USB and single-lane PCIe sound devices, not to mention SPI and I2C ADCs and DACs, use a 1-bit data bus. So the width of the data bus is of absolutely no consequence for the capabilities of the audio. If we're talking about 8-bit audio rather than 8-bit computers, the SK-1 (from 01985, two years later) shows you what you can do with 8-bit-wide digital audio. (It was also an 8-bit computer, but the reason its pianos, trumpets, and human voices sound like pianos, trumpets, and human voices is that they're sampled.) Or, say, Sprint's digital long-distance telephone service, launched in 01980 as United Telecommunications and turning a profit for the first time in 01989: entirely 8-bit mulaw audio. (That's not why telephone-quality voice sounds bad, though; that happens because of the sample rate, not the bit depth.) > Lastly the fact that you could copy software is really a big point on the + side. Piracy is the feature! Well, of course, but that's a feature of computers in general; it has nothing to do with 8-bitness. There's a continuing effort by bad people to prevent computers from being programmable, but it's mostly unsuccessful.

bullen 5y ago
Thx for clarifying things! You seem to like confusing dates, what is wrong with ISO 8601?
- kragen 5y ago
  Sure! ISO 8601 is not Y10K-compliant. :)
  - bullen 5y ago
    Why not? YYYYY-MM-DD works!
kragen 5y ago
> Every sound card on an IBM PC ISA bus also used an 8-bit data bus, including things like the Gravis UltraSound and the Sound Blaster AWE32. Erratum: the ISA bus had 16 data lines from the time of the IBM PC AT, and I believe that both the GUS and the AWE32 were 16-bit ISA cards, not 8-bit ISA cards.