1 comment

[ 0.19 ms ] story [ 10.0 ms ] thread
How much does printing to stdout in parallel tests skew the benchmark? Isn't stdout be definition synchronized and you are better off benchmarking without stdout involved?