There will be three standard sets of compiler flags, feel free to experiment with your own and let me know if you get anything faster. Always the tradeoff between accuracy and speed, so here are the basic standards with no debugging or tracing. You should do this in development before you release the code. These all relate to Athlon MP's or Xeon's.
Fastest speed, single precision.
Double precision with speed enhancements.
Double precision with no speed enhancements.
Single precision, fastest possible code.
1.
Double precision, with speed enhancements.
2.
Double precision, maintain accuracy at all costs.
3.
Single precision, fastest possible code.
1.
Double precision, with speed enhancements.
2.
Double precision, maintain accuracy at all costs.
3.
Single precision, fastest possible code.
1.
Double precision, with speed enhancements.
2.
Double precision, maintain accuracy at all costs.
3.
Single precision, fastest possible code.
1.
Double precision, with speed enhancements.
2.
Double precision, maintain accuracy at all costs.
3.
Single precision, fastest possible code.
1.
Double precision, with speed enhancements.
2.
Double precision, maintain accuracy at all costs.
3.
Single precision, fastest possible code.
1.
Double precision, with speed enhancements.
2.
Double precision, maintain accuracy at all costs.
3.
Some additional tweaks borrowed from RedHat.