Re: [Boost-users] [Signals] Another performance discussion

Subject: Re: [Boost-users] [Signals] Another performance discussion
From: Frank Mori Hess (frank.hess_at_[hidden])
Date: 2009-07-24 10:10:05

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

On Friday 24 July 2009, Sajjan Kalle wrote:
> What's surprising here is the long running time of the fragmented
> signal2 variant, which makes me belive there's a bug somewhere in it.
> Instead of connecting 100000 slots and disconnect 90000 randomly I
> connected 1000000 and disconnected 990000, the fragmented signal2
> performance on invocation then jumps up to about 92 seconds for
> calling a signal with 10000 slots.

That's due to the way signals2 cleans up disconnected to its slot list 
(incrementally during connection and invocation).  It only does a full 
cleanup when it has to do a deep copy of the slot list due to concurrent 
access.  If you run the signals2 fragmented block 10 times in a row, the run 
time will decrease as the disconnected slots are removed.  

However, I should be able improve the situation by adding a bit of code to 
invocation so it checks for an excessive number of disconnected slots and 
forces a full cleanup if needed.

> I think a lot of the disparities between signal and the vector variant
> actually is due to the underlying container used, a std::map will
> simply never beat the vector, and in fact signal will most likely
> perform considerably worse when slots go further and further apart and
> paging kicks in.

You should add a fragmented test using plain iteration over a std::list to 
your benchmark.  Also, you could try using a dummy_mutex for the signals2 
signals, like:

namespace bs2 = boost::signals2;
using bs2::keywords;
bs2::signal_type<void (void), mutex_type<bs2::dummy_mutex> >::type
  boostSignal2Fragmented, boostSignal2Unfragmented;

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.9 (GNU/Linux)

iEYEARECAAYFAkppwL0ACgkQ5vihyNWuA4VeMgCgoyLQhVVdg4j0iwReioWua1+I
/UYAn2MGVTsCz4qaDmGefT97aVS5ge3H
=eqyR
-----END PGP SIGNATURE-----