The Performance Cost To A Proposed Fedora 37 CFLAGS/CXXFLAGS Change

I loaded dozens of different open-source benchmarks/workloads for testing the impact of the "-fno-omit-frame-pointer" flag from open-source games to more technical computing workloads. It didn't take long for finding this option having some impact on the overall performance of the software being built with this flag -- just the software under test and not the entire OS rebuild as the F37 change proposal is encouraging.

In some of the tested workloads the addition of "-fno-omit-frame-pointer" did have very little impact on performance.

Meanwhile in the Botan crypto library is where this compiler flag led to the most significant performance regressions. Across various crypto algorithms tested, the Botan performance was destroyed by this compiler flag.