Do they ? If your usage is 50% userspace, 50% kernel, then having each slice slow by 10% does not result in a 20% slowdown.

They do... because they are measured in a weird way.

If the usage is split 50/50 kernel/userspace. And the userspace changes slow the app down 10%. Then the user space changes slowed the userspace code down 20%. Likewise if the kernel changes slow the app down 10%, they slowed the kernel code down by 20%.

So the total impact of the changes is 20% in your example.

That is indeed a weird way to measure and report. Without the associated userspace/kernel usage ratio those numbers are meaningless.

They aren't meaningless, they are measuring exactly what you care about. How much more computing power do you need to achieve the job.