Vecmathlib provides efficient, accurate, tunable, and most importantly vectorizable math functions such as sqrt, sin, or atan.
The library is implemented in C++, and intended to be called on SIMD vectors, e.g. those provided by SSE, AVX, or available in Power7 and Blue Gene architectures. The same algorithms should also work efficiently on accelerators such as GPUs. Even without vectorization, vecmathlib's algorithms are efficient on standard CPUs.