Page 1 of 1

Performance improvement opportunity?

Posted: Wed Nov 30, 2011 3:41 am
by Astara
Don't know if you'd be interested or how good it is, but I found this and it looks stable:

libev

Benchmarks against libevent 2 are rather interesting (sorry so big, but this site uses 'bb' code and it doesn't seem to allow one to size down an image like html does...

The Great thing about it -- it has a 'libevent' compatibility mode... that's almost as fast...

Have too look at the graphs a bit... but the per/loop processing time libevent2 (slightly slower than libevent), was 2.5-3x slower..
dat.t1.png
dat.t1.png (98.01 KiB) Viewed 2722 times

Re: Performance improvement opportunity?

Posted: Wed Nov 30, 2011 4:25 pm
by killemov
These graphs show promising improvement possibilities indeed. However, I could not find any license statement in a quick scan of schmorp.de. This means in many countries this code could suffer from the no-license license. This means it should be regarded as research material only. The original writer can slap on any license later on, even commercial ones. I do believe it is just an omission on the writer's part. I will contact him to make it LGPL, MIT BSD, ... whatever. As long as he makes it clear.

Re: Performance improvement opportunity?

Posted: Sun Jan 01, 2012 11:14 am
by killemov
It was clear, just had to read the LICENSE file from the source itself. :roll: