strange performance differences of 1.2.2 vs 1.2.5

[Chris Goffinet]

Excellent. I will try this on Monday and see where we go. I do believe  
we were using libevent 1.3e on FreeBSD (if not 1.3b for sure)

-- 
Chris Goffinet
MyBlogLog Senior Performance Engineer

Yahoo!
San Francisco, CA
United States

On Jun 1, 2008, at 4:30 PM, Sean Chittenden wrote:

>> That's a good guess (Brian? You know about that?:) ... but a bit  
>> doubtful.
>>
>> That gettimeofday(2) call that was changed is only made once per  
>> second
>> asyncronously. In order to have such a large performance impact  
>> we'd be
>> looking at some more broad changes.
>
> Hrm....  yeah, it's wrapped in an evtimer_set(3) call.  Wait...
>
> libevent(3's gettime() static function will use gettimeofday(2) if  
> either HAVE_CLOCK_GETTIME and CLOCK_MONOTONIC aren't defined and  
> use_monotonic is false.  Chris, I haven't profiled anything but it  
> may be worth looking into upgrading libevent(3) since it seems as  
> though the gettimeofday(2)/gettime() foo was still being toyed with  
> around the timeframe of libevent 1.3.  YMMV as this was only a  
> passing glance at the problem w/ the details provided.  :~)   
> gprof(1) memcached(8) if you want more specific guidance as to what  
> it could be.  -sc
>
>
> Stack:
>
> libevent(3)
> event_base_loop(3)
> timeout_correct()
> gettime()
>
>
> --
> Sean Chittenden
> sean at chittenden.org
>