c++ - Measuring performance/throughput of fast code ignoring processor speed? -
is there way write "tool" analyse produced x86 assembly language c/c++ program , measure performance in such way, wouldnt matter if ran on 1ghz or 3ghz processor?
i thinking more along lines of instruction throughput? how write such tool? possible?
i'm pretty sure has equivalent halting problem, in case can't done. things such branch prediction, memory accesses, , memory caching change performance irrespective of speed of cpu upon program run.
Comments
Post a Comment