It wasn't measured emperically as we're not very worried about that aspect. Still, the author of Ziglyph, jecolon, did some rudimentary comparisons and found it

It wasn't measured emperically as we're not very worried about that aspect. Still, the author of Ziglyph, jecolon, did some rudimentary comparisons and found it was equal and sometimes better by a few milliseconds. I suspect the size of the file being much smaller means in practice it now often fits in L1 cache during decompression, so that's probably counteracting the time spent decompressing. It also helps that the algorithm is able to rely on the files already being well sorted, making it a pretty linear process to compress.