2

I'd like to know how the CPU benchmark results for CPUs posted at, for example, http://www.cpubenchmark.net/high_end_cpus.html, compare. Is the scale linear, so that a CPU with a score of 2,000 is, on average, twice as fast as one with a score of 1,000?

Brian
  • 717
  • 2
  • 9
  • 14

1 Answers1

3

It seems to be linear; a 2000 score system completes the test units twice as fast as a 1000 score system.

The benchmarks seem to be completely artificial though, so in a real-world situation the ranks and numbers presented may be very skewed or even reposition.

Hyppy
  • 15,458
  • 1
  • 37
  • 59
  • Why do you think it's linear? Can you point me to a page? – Brian May 23 '11 at 19:52
  • Benchmarks traditionally use linear scales which is mostly due to the fact that most numbers represent some direct derivative of the "operations per timeunit" calculation. No idea, if the passmark calculations are published somewhere, but by the look of it they simply represent some [FLOPS](http://en.wikipedia.org/wiki/FLOPS)/[MIPS](http://en.wikipedia.org/wiki/Million_instructions_per_second) mixup – the-wabbit May 23 '11 at 20:27
  • @Brian they don't state this specifically on any page, however the relative speed of each CPU indicates a linear scale. If they said it somewhere then you wouldn't have to ask, now would you? ;-) – Hyppy May 24 '11 at 11:59