How do I interpret the benchmark results showing that it's slower than mergesort? Is that using the very simple implementation in vsort/examples /benchmark.c? Being slower than that while using SIMD and multicore isn't appealing.