industry_benchmarks uses a working set which fits into memory. This means operations performed in the benchmark don't incur a lot of disk I/O. Auditing forces operations to incur writes. If no I/O is otherwise performed by the test, this might overstate the total impact of using auditing.