Exploiting SIMD instructions in current processors to improve classical string algorithms