[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]

Re: Low benchmark performance



On Mon, 28 Apr 2003, Jeff Trawick wrote:

> Why does the test program use pthread_cond_broadcast() instead of
> pthread_cond_signal() to wake up a consumer?
The idea was just to see how it would perform with NPTL to wake all
threads at once. Probably a pointless and stupid test, but the HUGE performance
gap with linuxthreads seemed big enough to discuss it.

> What happens when you use pthread_cond_signal() instead?
NPTL rocks. It is (depending on the number of consumers) 2 to 6 times
faster than linuxthreads.

	Simon.





[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]