Is it true that heapsort is empirically just as fast as mergesort?

Empirically, heapsort and mergesort have similar performance in terms of speed, but the specific efficiency may vary depending on the data set and implementation.