1

I'm looking for a way to contextualize generational differences in computing performance. Is this a correct application of the MIPS measurement?

MOS 6502 - 0.500 MIPS at 1 MHz. As an 8-bit CPU it could process a theoretical maximum of 500,000 bytes a second.

Motorola 68000 - 0.700 MIPS at 8 MHz. As a 16-bit CPU it could process a theoretical maximum of 1,400,000 bytes a second.

Where am I going wrong here? I'd like to measure raw data throughput as I'd like to compare these integrated circuits to mainframes from the 1960s and 1970s.

Improve this question

1 Answer 1

Reset to default
0

It doesn't really make sense to use "bytes processed per second" as a metric. Even if it did, it's not even clear what "8-bit" means for the 6502 - it had an 8-bit accumulator and two 8-bit index registers, but a 16-bit address bus and multiple 16-bit addressing modes. Opcodes can be one or two bytes.

I am not as familiar with the 68000, but I know it's a RISC CPU - a rather different architecture, it has a lot more registers, and going by Wikipedia, 16 32-bit registers (8 general-purpose registers and 8 address registers, one of which served as the stack pointer, as opposed to the 6502's one accumulator, two index registers, a program counter, a stack pointer and flags register). The address bus was 24 bits (meaning it could address up to 16 MiB of memory instead of the 64 KiB available to the 6502).

The 68000 was much faster and more powerful than the 6502, but newer and more expensive as well. Unfortunately there is no perfect (or, arguably, even good) way to easily compare completely different processor architecures using one number. Attempts at constructing synthetic benchmarks like the Dhrystone score tend to be flawed in somre respect.

Improve this answer
2
  • Thanks for your answer. What about just comparing "integer performance" or "comparison operations?" Instructions per second isn't really ideal because more complex instructions can take longer to execute. But integers offer a baseline that I think is applicable to the type of performance - data sorting - that I'm trying to measure.
    – Schmudde
    Apr 8, 2014 at 15:40
  • This would be more reasonable. I still think you're going to have a hard time turning it into a single number that you can compare, though. Especially comparing a single chip to a mainframe - there are a lot of implementation details that will complicate the comparison.
    – user55325
    Apr 9, 2014 at 4:24

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .