Intel Xeon Cascade Lake Compiler Performance - GCC 9/10 vs. LLVM Clang 8/9
With the HPCG high performance computing benchmark, LLVM Clang produced slightly better performance over the GCC releases tested.
With FFTW for this FFT benchmark, the GCC compilers were still delivering better performance particularly for larger 2D FFTs.
With the HMmer molecular biology benchmark meanwhile LLVM Clang delivered better performance than GCC on this dual socket Intel Xeon server.
With the SciMark 2.0 micro-benchmarks, GCC 9/10 were performing slightly better than LLVM Clang but the 9.0 Git/SVN performance is inching closer into the GNU compiler territory.
LLVM Clang yielded faster performance than GCC for the TSCP chess benchmark but GCC 10.0 is in better shape after addressing a GCC 9 regression.
The performance flipped around yet again, for John The Ripper GCC was offering clearly better performance.