I have just downloaded a trial version of 32 bit version of intel c++ compiler for Pentium-4. I was experimenting with some simple linear algebraic stuff to see if it performs any better than the standard VisualStudio-6.0 compiler. I am curious whether any of you know whether it does any sort of paralellization when you use standard class "valarray". And if the answer is yes, how can I get it do it ?
I have tried several optimization options documented in the tutorial pages of the above-mentioned intel compiler but in a matrix-matrix multiplication I could not get any better result than the standard pointers. So valarray does not seem to get the stuff any faster in this compiler. Can anyone help me with this ?
valarray paralellized ?