I thought I posted the command in the readme, nmap -sP. Yeah I remember reading about nmap adding parallelization. According to the link you shared: "By default, Nmap calculates an ever-changing ideal parallelism based on network performance". So my benchmarks are based on comparing against that. I don't know why my implementation is faster, but the three tests I performed show that findpi is faster. I guess because I am forcing more threads than nmap's ever-changing ideal is allocating?