GCC vs. Clang Benchmark Comparison At Varying Optimization Levels
Under Himeno, GCC was faster than Clang at -O0 but Clang jumped out ahead at -O1, but then ended up losing out to GCC at the rest of the optimization levels tested.
While LLVM Clang used to compile code dramatically faster than GCC, the leads aren't as significant as they once were when making use of Clang 3.8 and GCC 5.3.1.
At -O2 and above is where GCC became unstoppable for C-Ray with Clang no longer offering comparable performance.
The FLAC encoding test started out with GCC being much faster than Clang when not dealing with optimizations, but by the time the test ended, GCC and Clang were offering comparable speeds.