Optimizing the OpenMP algorithm for computing all-pairs shortest paths on x86 architectures leads to significant performance improvements.