DESCRIPTION Conclusion: Testing on my system (IPC::System::Options 0.24, perl: 5.22.0, CPU: Intel(R) Core(TM) i5-2400 CPU @ 3.10GHz (4 cores)) shows the overhead to be ~40μs (0.04ms) so for benchmarking commands that have overhead in the range of 10-100ms we normally don't need to worry about the overhead of IPC::System::Option (0.04-0.4%) when we are using default precision (~1%). POD ERRORS Hey! The above document had some coding errors, which are explained below: Around line 5: Non-ASCII character seen before =encoding in '~40μs'. Assuming UTF-8