Some benchmarks used the Walltime instrument outside CodSpeed Macro Runners, which can make the results inconsistent.
This benchmark measured no execution time, so its result is not comparable and counts as unchanged. This usually happens when the compiler removes the code under test.