GCC 9 vs. Clang 8 C/C++ Compiler Performance On AMD Threadripper, Intel Core i9
With the FFTW benchmark, GCC 9.1 did yield a faster binary on the Core i9 7980XE compared to GCC 8.3 but that was the opposite case on the Threadripper system. For the Intel system the faster GCC 9.1 performance allowed it to pull out further ahead of Clang 8 while the 2990WX saw Clang performing between the performance level of GCC 8 and GCC 9.
MAFFT was slightly faster on GCC 9 and let it perform in line with Clang 8.
For the SciMark2 C language micro-benchmarks, the Intel Skylake-AVX512 performance saw GCC 9.1 roughly the same as GCC 8 and in turn faster than Clang 8. The AMD Ryzen Threadripper box meanwhile saw GCC 9.1 running faster and that put it to par with the Clang 8 performance.
One workload where Clang was performing much faster than GCC on both systems was with the TSCP chess benchmark.
Clang also appeared to perform much better with the Intel MKL-DNN math kernel libraries for deep learning.
Clang also was performing well for Intel's SVT-AV1 video encoder.