I think flameduck meant code cache not data cache...
and I think he has a very good point.
The rather minimal total amount of time you will save on certain machines just aint worth the hassle.
Far better to write somthing that actually works in the real world, and then get *actual usage timings* for various routines.
Whats far more likly to help is refining the algorithm you are using or even completely changing how you are do things, which will often save much more execution time.
and I think he has a very good point.
The rather minimal total amount of time you will save on certain machines just aint worth the hassle.
Far better to write somthing that actually works in the real world, and then get *actual usage timings* for various routines.
Whats far more likly to help is refining the algorithm you are using or even completely changing how you are do things, which will often save much more execution time.