There are a few interesting areas where you fall slightly short of the other benchmark you cite. In almost all, you beat the other benchmark. Both are running on Mac Pro 5,1 systems. You are running 10.11, the other benchmark is running 10.10. You have all four memory slots populated, which is known to slow memory access by as much as ten percent in round numbers. The other benchmark is on a Mac with the 3 populated slots optimally interleaved.
In Integer perfomance, only a few very complex multiple-core operations such as AES (encryption instructions), Sobel, Lua, Djistra fall short of the other benchmark -- all others are faster.
In Floating point operations, all of yours are faster.
In memory access, single-core are slightly slower, multi-core are quite a bit slower.
It is possible the tests have not been constructed as a level playing field for the newer processor, or that changes in 10.11 system libraries have slowed some operations down. The memory difference may be mostly down to your use of the fourth shared memory slot. I would certainly ask the test developer about those differences.
Overall, I would say the performance of the two systems on this test are roughly comparable.
