[dtrace-discuss] dtrace overhead and proper measuring technique

[Joel Reymont]

On Sep 9, 2009, at 9:58 PM, James McIlree wrote:

> Given that you have only 40 hits, my guess is you should use the .9s  
> context.

May I ask why?

Can this be generalized into saying that function run time as measured  
by timestamp should _always_ be evaluated against total eval time  
measured by excluding the function probe?

I still don't understand whether nsMenuX::MenuConstruct really takes  
12ms of elapsed time as DTrace is claiming. I will try to measure this  
using mach_absolute_time, without involving dtrace, to see if it makes  
a difference.

	Thanks, Joel

---
fastest mac firefox!
http://wagerlabs.com