What optimization can be made for nanosecond IO and CPU stability when performing a timing attack?

What optimization can be made for nanosecond IO and CPU stability when performing a timing attack?

I'm using Rust to create a program to attempt a timing attack on a network resource (a printer I lost a password to). I'm wired directly into it. What Linux environmental constraints can I optimize to minimize noise and variability?

Currently, even over 100,000 runs on a single threaded process I do not see a statistic significance.