I avoid VMs when benchmarking for the same reasons you cite. Benchmarking is difficult enough to get reproducible results without adding another variable to an

I avoid VMs when benchmarking for the same reasons you cite. Benchmarking is difficult enough to get reproducible results without adding another variable to an already noisy process.