GCC vs. Clang Compiler Performance On NVIDIA Xavier's Carmel ARMv8 Cores
Within the basic Scimark C micro-benchmarks, the GCC compiler tended to have a minor to rather significant benefit over Clang with the exception of the Jacobi successive over-relaxation sub-test where Clang was about 6% faster.
With the Himeno pressure solver, the results were virtually tied.
Additionally, with the asmFish chess benchmark the results came in as a tie albeit not surprising since the performance sensitive code in asmFish is hand-tuned Assembly.