0

Should the same program always output the same result, if it does not use any random numbers or I/O, or is it possible that it will output something different on a different cpu (but same architecture, no need for recompile) ? I'm specially thinking about calculations with floats and doubles which depend heavilly on precision, as used in numerical approximations.

miguel.negrao
  • 949
  • 5
  • 13
  • 1
    The question http://stackoverflow.com/questions/968435/what-could-cause-a-deterministic-process-to-generate-floating-point-errors?rq=1 seems to give a negative answer to my question. – miguel.negrao Apr 10 '13 at 23:01

1 Answers1

0

I don't believe it should ever happen, since CPUs using the same architecture should basically use the same set of registers that they use for storing data.

Short: The result will be the same.

Nitkov
  • 468
  • 6
  • 18