Testing stockfish from git compiled with latest released compilers under OS X 10.10.4 on an overclocked Ivy Bridge 3770K running at 4.7Ghz.
for i in stockfish-*; do echo -n "$i :" ; for j in $(seq 1 5); do ./$i bench 128 4 18 ; done 2>&1 | awk '/Nodes\/second/ {sum+=$3} END {printf(" %.0fkn/s\n", sum/5000)}'; done
stockfish-4.7 : 9386kn/s
stockfish-4.8 : 9151kn/s
stockfish-4.9 : 9388kn/s
stockfish-5.1 : 9590kn/s
stockfish-5.2 : 9665kn/s
stockfish-clang-3.6.2 : 8703kn/s
stockfish-intel : 9197kn/s
stockfish-osx : 8660kn/s
Clearly the gcc folks are doing a great job and keep improving with each release.