LLVM Clang 3.9 Mostly Trails GCC In Compiler Performance
With the Fhourstones benchmark, Clang 3.8.0 was the fastest compiler benchmarked.
In the SciMark2 micro-benchmarks is where Clang tended to perform much better than GCC.
In some sub-tests of SciMark2 the Clang performance is much slower...
Or close to the same...
But with the Sparse Matrix Multiply and Dense LU Matrix Factorization tests is where the LLVM compiler sped ahead, which boosted its composite score shown earlier.
The Jacobi test also heavily favored Clang.