yade-users team mailing list archive

[Janek Kozicki <janek_listy@xxxxx>]

Janek Kozicki said:     (by the date of Thu, 15 Sep 2005 16:34:09 +0200)

> I have noticed that new version of boost is released - 1.33. Boost
> authors claim that they have made some improvements in the code.
> 
> So I have done some benchmarks, and indeed - when yade is compiled with
> boost 1.33 it is 30% faster. Test calculation with boost 1.32 takes 100
> seconds, and with boost 1.33 takes 70 seconds.
> 
> so if you need faster calculations - use boost 1.33 :)

oh, wait, boost 1.33 is not suitable for latest release
yade-all-0.9.0.911, there are some compilation problems in
ErrorTolerant. to fix it - simply replace all sparse_matrix with matrix,
and change those lines in ErrorTolerantLaw.cpp:

ublas::sparse_matrix<float> J (6*bodies->size(), ncb->volatileInteractions->size(), ncb->volatileInteractions->size()*2/*6*bodies.size()*body->volatileInteractions.size()*/);
ublas::sparse_matrix<float> Jt (ncb->volatileInteractions->size(), 6*bodies->size(), ncb->volatileInteractions->size()*2/*6*bodies.size()*body->volatileInteractions.size()*/);

to

ublas::matrix<float> J;
ublas::matrix<float> Jt;

this will break ErrorTolerant and make it compilable with boost 1.33,
but I hope that Olivier will soon fix it.

If you don't want to do such modifications - use boost 1.32.

Next release should be fixed for boost 1.33 ;)

-- 
Janek Kozicki                                                         |
_______________________________________________
Yade-users mailing list
Yade-users@xxxxxxxxxxxxxxxx
http://lists.berlios.de/mailman/listinfo/yade-users