pmeerw's blog

27 Mar 2010

Sat, 27 Mar 2010

Floating-point arithmetic performance

Sometimes you really want to know... (continuation of my previous writeup)

Runtime for various double-precision floating-point arithmetic operations, normalized by clock frequency, across several CPU models; P4 and Opteron suck, E8400 is consistently the fastest. Surprisingly, div and floor are quite slow, pow is dead slow.

posted at: 18:10 | path: /programming | permanent link

Made with PyBlosxom