For the following performance options, those with the greatest significance are listed first.
Table 3-8 Performance Options
Action |
Option |
Details |
---|---|---|
Faster execution and compilation on native platform by using a set of options. | -fast | |
Optimize performance--set optimization level | -O[n] | |
Specify target hardware system. | -xtarget=t | |
Collect or use data for a profile to optimize (SPARC). | -xprofile=p | |
Align COMMON block data and enable multi-word load/stores. (SPARC) | -dalign | |
Arithmetic--use simple arithmetic model. | -fsimple | |
Arithmetic--use SPARC non-standard floating point (SPARC). | -fns | |
Use inline library | -libmil | |
Traps--assume no memory-based traps (SPARC). | -xsafe=mem | |
Unroll loops--advise optimizer to unroll loops n times. | -unroll=n | |
Fast math--use special fast math routines (SPARC). | -xlibmopt | |
Architecture--specify target instruction set. | -xarch=a | |
Chip--specify target processor. | -xchip=c | |
Check data dependencies--analyze loops (SPARC). | -depend | |
Inline the listed user routines to optimize for speed. | -inline=rlst | |
Optimize across all source files on command line | -xcrossfile | |
Do no optimizations that increase code size. | -xspace |