Different hashing algorithm. As already noted, you're only testing insertion, so the only difference will be due to how the value is hashed, with some smaller difference attributable to the precise structure of the hash bins.

If you want more details, you'll need the code for both algorithms. Honestly, I don't see much difference. Are you really going to care if a million inserts takes 0.05 seconds longer? Particularly when it's much more likely to be retrieval rather than insertion speed which is more important?