A proposal for autotuning linear algebra routines on multicore platforms

The use of an OpenMP compiler optimized for the corresponding multicore system is a good option, but it is possible in a system to have access to more than one compiler and different compilers can appropriately optimize different parts of the code. In this paper we present a proposal for an autotuning system for linear algebra routines that decides the best compiler for each situation, as well as other parameter values, as, for example, the number of threads to generate.