05-26-2010, 10:48 PM
This is really impressive!
On the other hand, I would like to benchmark your fast FP routines. If they really improve speed we should consider to integrate them directly into the compiler (e.g. using a --fast-floating-point command line flag as other compilers do).
What do you think?
On the other hand, I would like to benchmark your fast FP routines. If they really improve speed we should consider to integrate them directly into the compiler (e.g. using a --fast-floating-point command line flag as other compilers do).
What do you think?