Here is a blackfire comparison of the loading of the fixture file used by the benchmark (I haven't profiled the whole benchmark, as repeating it 100 times leads to some functions being called thousands of millions of time, which makes the overhead of the profiler increase too much): https://blackfire.io/profiles/compare/0ad49ded-25b3-483b-aacd-3ccbc203db69/graph
Here is a blackfire comparison of the loading of the fixture file used by the benchmark (I haven't profiled the whole benchmark, as repeating it 100 times leads to some functions being called thousands of millions of time, which makes the overhead of the profiler increase too much): https://blackfire.io/profiles/compare/0ad49ded-25b3-483b-aacd-3ccbc203db69/graph