[Author Prev][Author Next][Thread Prev][Thread Next][Author Index][Thread Index]
Re: [Libevent-users] Hacks for getting around CONFIG_HZ
On 8 Oct, 2009, at 10:58 , Florian Weimer wrote:
I know this is unlikely, but do you know any hack to get around this?
Use a tickless kernel?
Thanks for your answer Florian, that's a good suggestion but before
doing that (or even just increasing CONFIG_HZ to 1000) I was looking
for some libevent hack to obtain the same result.
On 8 Oct, 2009, at 12:21 , Eric M. Hopper wrote:
if you
are doing IO driven events that you can end up with an event that is
fired many more times per second
Luckily yes, that's just for timers.
Why, exactly, do you need an event that will fire more than 250
times a
second? What are you using it for? Will polling work instead?
In this specific case, I'm running some network benchmarks where a
sender process sends data at a fixed rate.
To do that I was trying to wake up every millisecond and call send() N
times.
For the moment i resolved by waking up every 4 milliseconds and
calling send() N*4 times,
not the same but fair enough for this purpose.
***********************************************************************
To unsubscribe, send an e-mail to majordomo@xxxxxxxxxxxxx with
unsubscribe libevent-users in the body.