[Carpet] Carpet asserts

Erik Schnetter schnetter at cct.lsu.edu
Fri May 26 17:35:40 CEST 2006


On May 26, 2006, at 06:06:40, Wolfgang Kastaun wrote:

> Hi !
> Suddenly I got these assertions:
>
> cactus_tov_opteron: /net/hydra/home/kastaun/wd/cactus/latest/ 
> configs/tov_opteron/build/
> Carpet/Recompose.cc:763: void Carpet::SplitRegions_Automatic(const  
> cGH*, std::vector<ib
> box, std::allocator<ibbox> >&, std::vector<bbvect,  
> std::allocator<bbvect> >&, std::vect
> or<int, std::allocator<int> >&): Assertion `abs(prod(rshape) -  
> nprocs) < 1e-6' failed.
> cactus_tov_opteron: /net/hydra/home/kastaun/wd/cactus/latest/ 
> configs/tov_opteron/build/
> Carpet/Recompose.cc:763: void Carpet::SplitRegions_Automatic(const  
> cGH*, std::vector<ib
> box, std::allocator<ibbox> >&, std::vector<bbvect,  
> std::allocator<bbvect> >&, std::vect
> or<int, std::allocator<int> >&): Assertion `abs(prod(rshape) -  
> nprocs) < 1e-6' failed.
>
> What does it mean ?
>
> This was a 8-cpu run, the executable worked fine before in an  
> almost identical setup
> (only the resolution did change).

These lines check whether the processor decomposition is consistent.   
The result should be integer, but the calculation is done using  
floating point numbers, and the fudge factor 1.0e-6 is supposed to  
allow for floating point roundoff.  This check is not supposed to  
fail for any user input.

It would be interesting to know the specific values of these  
variables.  Can you print them to the screen, or in a debugger, or  
send me the parameter file (preferably with bugzilla)?

-erik

-- 
Erik Schnetter <schnetter at cct.lsu.edu>

My email is as private as my paper mail.  I therefore support encrypting
and signing email messages.  Get my PGP key from www.keyserver.net.



-------------- next part --------------
A non-text attachment was scrubbed...
Name: PGP.sig
Type: application/pgp-signature
Size: 186 bytes
Desc: This is a digitally signed message part
Url : /archives/developers/attachments/20060526/942ad281/attachment.pgp 


More information about the developers mailing list