For years DragonFlyBSD has been working on multi-core improvements, largely by Matthew Dillon, the leader of this FreeBSD fork. We have looked at the BSD CPU scaling performance in the past but since then they have continued to work on performance improvements.
The latest work by Matthew Dillon is on reducing SMP contention with having worked out a name-cache shared lock fix, greater use of shared spinlocks in pmap, a file-system syncer improvement, fork and fork/exec code-paths are more efficient with much less lock contention, the per-CPU process reaper now uses a per-CPU token rather than a global token, and PID-related improvements.
For multi-core systems, the DragonFlyBSD performance should be a lot better for process-related activities, particular those relying upon fork/exec like in the make command or shell scripts.
Matthew Dillon ended his exciting announcement on the DragonFlyBSD mailing list with "I consider this a fairly major milestone for the project. We aren't finished, but this is a major leap in our ability to fully utilize the resources on larger multi-core systems."