I once (8 years ago) did a comparison between random, pseudorandom and quasirandom. Quasirandom worked ever so slightly better than the other two in speed and final performance, but not enough to warrant the additional complexity of implementation.

It is pseudoworse.