Benchmarking AMD's New AOCC Compiler For Ryzen
More benchmarks where AOCC is running right in line with Clang...
Finally with the AVX-heavy Himeno pressure solver benchmark we see a test case where AOCC 1.0 does appear to deliver slightly better performance than LLVM Clang 4.0: a gain of about 3%.
With having more optimizations, AOCC 1.0 is the slowest compiler tested at the compilation process of C/C++ codes.