AmpereOne Performance With GCC vs. LLVM Clang Compilers

While having the Supermicro ARS-211M-NR R13SPD server in the lab for AmpereOne benchmarking with the flagship AmpereOne A192-32X processor, I took the opportunity to run some fresh GCC vs. LLVM Clang compiler performance benchmarks on AArch64. Here are those results for that healthy competition between these open-source C/C++ compilers on AmpereOne cores.

This round of testing was using GCC 13.2 as shipped by Ubuntu 24.04 LTS compared to using LLVM Clang 18.1.3 as available via the Ubuntu 24.04 archive. Plus using LLVM Clang 19.1 and Clang 20.0 Git using the Debian/APT upstream packages from LLVM.org. I didn't do any other combinations like going for GCC 14 / GCC 15 development as primarily was interested in seeing the Clang performance and had to limited the tested combinations due to only having this Supermicro AmpereOne server in the lab for a few weeks before needing to return it.

There were no changes to the system hardware/software besides swapping out the compiler being used for conducting these C/C++ open-source benchmarks. The CFLAGS/CXXFLAGS were kept the same throughout testing and set to "-O3 -march=ampere1 -flto" for an O3 optimized look and using Link Time Optimizations on all the compilers tested.

Let's continue on with this fresh AArch64 compiler comparison quickie.