yade-dev team mailing list archive

[Anton Gladky <gladky.anton@xxxxxxxxx>]

Hi, Vaclav,

I will send you more simple script, which makes also a problem.
Machines are: Intel E5200, Intel i5, Intel i7, AMD Phenom X6.

Only i7 makes a problem. And, as I told you, Maverick on VirtualBox (i386)
also gives this error.

Anton



On Tue, Sep 7, 2010 at 1:31 PM, Vaclav Smilauer <eudoxos@xxxxxxxx> wrote:

>
> > Hi, Vaclav!
> >
> > It seems, I have found the commit, which makes me problem on one of my
> > machine.
> > It is a very strange situation. I have 4 machines for simulation. And
> > only one of them react on this problem.
> Hi Anton, it seems odd indeed. In that particular revision, there were
> no big changes, except that BoundDispatcher was put inside the Collider
> and Bo1 functors were turned into real Bo1 functors (creating
> Body::bound for each body where called rather than just updating it). If
> you can give the script (you can send it to me by private mail if it is
> confidential), that would help a lot.
>
> Are those machines different in architecture? Intel CPUs notoriously
> have issues with denormalized values (nans and infs), but they should
> not appear in the simulation anyway.
>
> Cheers, v.
>
>
>
> _______________________________________________
> Mailing list: https://launchpad.net/~yade-dev<https://launchpad.net/%7Eyade-dev>
> Post to     : yade-dev@xxxxxxxxxxxxxxxxxxx
> Unsubscribe : https://launchpad.net/~yade-dev<https://launchpad.net/%7Eyade-dev>
> More help   : https://help.launchpad.net/ListHelp
>