Hi, has anyone any experience using this?

It seems to be in some repos, but source code is also available, so it can be build anywhere:
http://jnovy.fedorapeople.org/pxz/

There are some benchmarks available, but I'd like to hear from some casual users how it perform, and if it is actually efficiently programed.

It adds two new switches to xz:
-T maximum number of threads to run simultaneously
-D per-thread compression context size as a multiple of dictionary size

Cheers,
--polemon