Every published test that I have ever seen shows Linux systems beating out FreeBSD systems in at least 70% of the tests. There are a few where FreeBSD does well, but there are a few others where Linux systems, especially Ubuntu, really leave FreeBSD in the dust.
In my own usage of Linux and BSD systems, both run fine, but if you run a GUI and use them for interactive use, Linux systems are more convenient to install and without a doubt faster in feel, which supports the benchmark research.
BSD based systems are lauded as great servers, but Red Hat Enterprise Linux has great support, which makes it a winner in the enterprise. Based on the tests here, consistently finding CentOS to be faster in the majority of tests, I don't see how we can come to any other conclusion than that the majority of Linux systems utilize system resources more efficiently than the majority of BSD based systems.
I first saw this in 1995 when a 100 MHz Micron PC running Slackware booted and started a light desktop environment faster than a 200 MHz AlphaStation 200 with more memory, bigger and faster disks, and generally more capabilities. I am sure that the Alpha would toast that old Micron on compiler performance and floating point arithmetic, but nevertheless, the inexpensive Micron system with Slackware felt more responsive to the touch - noticeably so, compared to one of the fastest workstations of that era. More seat of the pants evidence of something I noticed more than a decade ago.
More recently, my Debian systems boot to the desktop and respond more rapidly to routine input in a GUI than the very nice, but so-so performing PC-BSD. Nothing empirical here, just seat of the pants observation. But coupled with these benchmarks, I'd say the evidence is pretty strong: Ubuntu beats FreeBSD, Ubuntu beats PC-BSD (an earlier study here I believe), CentOS beats FreeBSD (as discussed in this thread), Slackware beats BSD UNIX in the nineties in response and feel, Debian beats PC-BSD is response and feel; subjective but observed. That seems to be enough evidence, especially without much contradicting this information, first observed, then empirically confirmed.