I never hear anything about 16-bit CPUs other than the 68000. Was there any other 16-bit CPU that was good for arcade games? How does their performance compare with the 68000?
Printable View
I never hear anything about 16-bit CPUs other than the 68000. Was there any other 16-bit CPU that was good for arcade games? How does their performance compare with the 68000?
The CPU in the Intellivision was 16-bit, Wiki says it is a "General Instrument CP1610 16-bit microprocessor CPU running at 894.886 kHz".
Less than one MHz vs. the 68000 in its variety of MHz flavors.
Also:
http://en.wikipedia.org/wiki/16-bit
The 68000 isn't even a true 16-bit processor. It has 32-bit registers.
Other 16-bit CPUs:
65816/65c816 (Apple IIgs, SNES)
8086 and 80286 (IBM-compatible PC's)
For arcade games though? All I could think of would be the SNES-based boards that had the same processor in them. The 68000 was pretty monolithic in its time.
Yeah the 68000 was essentially a 32bit CPU limited by 16bit data throughput. So any software developed for a 68000 machine would also work on the fully 32bit 68020, 68030, 68040 and 68060 and run faster.
The 68k cpu was used because it was widely used; plenty of people knew how to program it, it was cheap because loads were made, and there was a clear upgrade path if you needed a faster CPU.
I wonder what the 32x could have been like if it had used a single 68060 cpu instead of the dual Sh2 setup it had. It would have made it a hell of a lot easier to develop for, so there could have been a whole lot more games.
Actually, a little less simple than that. It only had 16bit ALUs. The ALU is what determines the CPUs bit-ness.Quote:
Yeah the 68000 was essentially a 32bit CPU limited by 16bit data throughput.
Other than the 65xxx, 8086/80186/80286, some other obscure processors. Looking at MAME source, the vast majority used 68k.Quote:
I never hear anything about 16-bit CPUs other than the 68000.
As Tomaitheous said, Arithmetic-Logic Unit is what tell the bit count. 68000 and 68010 are 16 bit as they have 16-bit ALUs, all later ones are 32bit as their ALUs are 32bit also. 68000 feels 32bit, but actually isn't it. I love 68000 because of that.
No, it's true. It has 32bit registers and has opcodes to handle 32bit arithmetic, but that doesn't make it a 32bit CPU. Like TmEE said, it definitely feels like a 32bit CPU. There are 8bit processors with 16 registers and 16bit arithmetic operations, but they aren't 16bit processors. There's even an 8bit processor with a 32bit register for 32bit arithmetic operations - still an 8bit processor. What makes the original 68000 16bit is the ALU. Just like 8bit CPUs are 8bit because of the ALU. The BUS doesn't define a CPU's 'bitness' either, but it is a measure of performance.
Motorola was ahead of the times when they released the original 68k in 1979. It was a forward compatible model that scaled nicely(almost seemlessly) into the 32bit revisions of the 68k family. There's no doubt that the processor was designed with high level language optimization in mind. I'll admit the 68k gets more credit about it being a power house than it deserves, though. I wouldn't doubt that came from the Amiga community. The original 68k has some slow operation times. It was more to the fact that the processor was available in 8, 12, and 16mhz VS the 2mhz 8bit CPUs at the time (the z80 @ 4mhz is roughly equiv to 1mhz against the 65C02 and 68B90E). Still, with the linear memory layout and nice fat set of registers, the 68k is lovely to code on.