Hi Steven,

| 1) Replaced the gettimeofday() call with 'rdtsc' instruction to avoid a
| kernel transition in the fast path.
I had similar problems with the delay_loop which also goes back to use
gettimeofday() (in a spin-loop). However, the rdtsc instruction is afaik
not portable, since peculiar to Intel CPUs. 

For this reason, the patch I submitted uses nanosleep() - on Linux
systems, this means using the hrtimers API, which internally also uses
the timestamp counter (or other available high-resolution counters), but
is more portable.

Gerrit


The University of Aberdeen is a charity registered in Scotland, No SC013683.


-------------------------------------------------------------------------
This SF.net email is sponsored by the 2008 JavaOne(SM) Conference 
Don't miss this year's exciting event. There's still time to save $100. 
Use priority code J8TL2D2. 
http://ad.doubleclick.net/clk;198757673;13503038;p?http://java.sun.com/javaone
_______________________________________________
Iperf-users mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/iperf-users

Reply via email to