Read the section of the provided Optimizing Applications manual on parallel processing. You can use the compiler's auto-parallel feature (best when combined with generating optimization reports to see what blocks the compiler from parallelizing loops) or, for better overall parallelization but more coding effort, OpenMP.
You should also take advantage of the compiler's other optimization features, such as Whole Program Optimization (/Qipo), Profile-Guided Optimization, /O3 and use the CPU-specific switch combination "/QaxT /QxW". This will run on any PC that has a Pentium 4 (or compatible) or later processor, but will optimize best for Intel Core 2 processors.
Depending on your application needs, look also at what the Intel Math Kernel Library has to offer.
