Subject: Re: [boost] boost::signal poor performance
From: Frank Mori Hess (frank.hess_at_[hidden])
Date: 2008-09-24 08:57:41
-----BEGIN PGP SIGNED MESSAGE-----
On Wednesday 24 September 2008 01:30 am, Hansi wrote:
> I have tested already also a vector<boost::function> against the
> boost::signal. The vector (able to disconnect during events) was about 2
> times slower than boost::function.
> I have made now also tests with the new boost::signal library which
> should replace in the future this one. It is about 70times slower....
> In the next days I want to make a comparsion with other eventhandlers..
> I will post than the results.
When I compile your benchmark under linux with g++ 4.1.2 and -O3, I get:
function 1.13 sec
signals 16.28 sec
thread_safe_signals 12.31 sec
Maybe you need to try some different compiler switches?
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.6 (GNU/Linux)
-----END PGP SIGNATURE-----
Boost list run by bdawes at acm.org, gregod at cs.rpi.edu, cpdaniel at pacbell.net, john at johnmaddock.co.uk