LLVM Clang 3.7 vs. GCC Compiler Benchmarks On Linux
Clang 3.7 expands it's lead slightly with the SciMark 2 C benchmark over Clang 3.6, which was already beating out GCC.
Partially responsible for the elevated composite number was the Sparse Matrix Multiply sub-test, which went from being slower than GCC to now being faster.
Clang also remains much faster with SciMarj's Dense LU Matrix Factorization and Jacobi Successive Over-Relaxation tests.