Hello, Wolfgang.

I looked at the memory consumption of the system_matrix, the system_rhs and 
solution vectors. The difference appears, as Timo suggested, in 
the system_matrix.

The difference may be that the NEW version uses a DynamicSparsityPattern 
while the OLD only guesses on the size 
with, dof_handler.max_couplings_between_dofs(). Apparently with 3d problems 
the function overestimates. Presently DynamicSparsityPattern is used in 
step-8, but, not in step-17.

Thanks

Pete Griffin

=====================================================================

NEW Code

Cycle 4:
   Number of active cells:       7484
dof_handler.n_dofs() 29277
solution.size() 29277
system_rhs.size() 29277
system_rhs.local_size() 29277
system_matrix.memory_consumption() 18320628 -> 18 MB
   Number of degrees of freedom: 29277 (by partition: 29277)
   Solver converged in 48 iterations.
   Peak virtual memory: 832 MB, Peak resident memory: 102 MB

=====================================================================

OLD Code

Cycle 4:
   Number of active cells:       7484
n_local_dofs 29277
solution.local_size() 29277
dof_handler.n_dofs() 29277
solution.size() 29277
system_rhs.size() 29277
system_rhs.local_size() 29277
system_matrix.memory_consumption() 361866548 -> 361 MB
   Number of degrees of freedom: 29277 (by partition: 29277)
   Solver converged in 48 iterations.
   Peak virtual memory: 1142 MB, Peak resident memory: 414 MB


On Sunday, July 31, 2016 at 12:52:11 AM UTC-4, bangerth wrote:
>
> On 07/30/2016 05:08 AM, Pete Griffin wrote: 
> > I used the methodology of step-18 for step-17 which showed memory usage 
> > improvements. I have attached a new version of step-17 that allow 
> selecting 
> > between the NEW and OLD. I also attached the results from both and 
> another 
> > plot. I don't know whether this version will work with more than one 
> processor 
> > I run only on 1. It might be helpful if someone would try to run it and 
> report 
> > success or failure. 
>
> Pete -- the two versions are quite different. Have you found out what 
> *exactly* it is that makes the difference in memory consumption? I would 
> like 
> to add a comment to this effect to the program. 
>
> Best 
>   W. 
>
> -- 
> ------------------------------------------------------------------------ 
> Wolfgang Bangerth               email:            [email protected] 
> <javascript:> 
>                                  www: http://www.math.tamu.edu/~bangerth/ 
>
>

-- 
The deal.II project is located at http://www.dealii.org/
For mailing list/forum options, see 
https://groups.google.com/d/forum/dealii?hl=en
--- 
You received this message because you are subscribed to the Google Groups 
"deal.II User Group" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to [email protected].
For more options, visit https://groups.google.com/d/optout.

Reply via email to