Versions Compared

Key

  • This line was added.
  • This line was removed.
  • Formatting was changed.

...

A Complete Example for a lossless cavity

...

Code Block

************************************************************

...

*

...

***

...

	Omega3P V8.0.0 07/01/2009 $

...

	**

...

*

...

------------------------------------------------------------

...

-

...

 Copyright 2009, Advanced Computations Department 

...

 SLAC National Accelerator Laboratory

...

************************************************************

...

*


Read Mesh: gun-v4.ncdf
Partitioning Method: parmetis

...

      Setup: Max:

...

   0.007, Sum:

...

   0.015, Balance:

...

   1.002

...

   Matching: Max:

...

   0.008, Sum:

...

   0.015, Balance:

...

   1.001
Contraction: Max:

...

   0.011, Sum:

...

   0.022, Balance:

...

   1.001

...

   InitPart: Max:

...

   0.001, Sum:

...

   0.001, Balance:

...

   1.000

...

    Project: Max:

...

   0.000, Sum:

...

   0.001, Balance:

...

   1.000

...

 Initialize: Max:

...

   0.002, Sum:

...

   0.004, Balance:

...

   1.002

...

      K-way: Max:

...

   0.006, Sum:

...

   0.011, Balance:

...

   1.000

...

      Remap: Max:

...

   0.000, Sum:

...

   0.000, Balance:

...

   1.000

...

      Total: Max:

...

   0.035, Sum:

...

   0.070, Balance:

...

   1.000

...


**********************************************************

...

*

...

*	Total Number of Elements read: 

...

	24868

...

*	Total Number of Elements used: 

...

	24868

...

*	Total Number of DOFs: 

...

	152902

...

**********************************************************

...

*

Total Volume of the strucutre is : 0.0001593555890525991
Number of Grad DOFs: 27613

...

*********************************************************

...

*

...

 ARPACK Loop: 
Shift = 439.2566356039645

...

*********************************************************

...

*
factorizing the matrix using MUMPS ...
Using ParMETIS for ordering...


Analysis step: 0.761027 seconds

...


	Maximal per-core estimated memory 

...

	 451 MB

...

	Aggregated estimated memory 

...

	 875 MB

...

	Maximal per-core estimated memory if OOC 

...

	 184 MB

...

	Aggregated estimated memory if OOC 

...

	 365 MB

Factorization step: 39.184121 seconds

...


	 ncv=6

...

	 nev=2

...

 Number of converged eigenpairs = 2
eigenvalue:

...

  2.726061596672462e+03 

...

	 Frequency:  2.491200411267457e+09 

...

	 Residual:   2.54e-11
eigenvalue:

...

  3.545447315320415e+03 

...

	 Frequency:  2.841033486606133e+09 

...

	 Residual:   1.25e-09
COMMIT MODE: 0 FREQ = 2491200411.267457

...

	 k= 52.21169980638881

...

	norm(v

...

[0

...

]) = 31.04704308504932
COMMIT MODE: 1 FREQ = 2841033486.606133

...

	 k= 59.54365890101494

...

	norm(v

...

[1

...

]) = 33.84520753989371
Number of TriSolve: 39. Average time for one TriSolv: 0.287939
Computed Total Energy (normalized by Epsilon0/2): 0.999999999999992
Computed Total Energy (normalized by Epsilon0/2): 0.9999999999999895

Some Explanations

The following lines show the number of elements and number of DOFs in the computation: 

Code Block

  ***********************************************************

...


  *	Total Number of Elements read: 	24868
  *	Total Number of Elements used: 	24868
  *	Total Number of DOFs: 	152902
  ***********************************************************

The following lines show the memory usage:

Code Block

...


	Maximal per-core estimated

...

 memory 	 451 MB
	Aggregated estimated memory 	 875 MB
	Maximal per-core estimated memory if

...

 OOC 	 184 MB
	Aggregated estimated memory if OOC 	 365 MB

If the Maximal per-core memory is larger than what is available, user should either increase the number of cores in the computation or use other options such as Out-of-core (OOC) solver. Note that it may not be good to use excessively large number of cores in the computation (in fact, it may hurt performance or it may fail to get results if doing so). It is often a good idea to have a few thousand elements per core in the parallel computations.

The following lines show the resulting eigen frequencies. Number of converged eigenpairs = 2
eigenvalue: 

Code Block

 Number of converged eigenpairs = 2
eigenvalue:  2.726061596672462e+03

...

 	 Frequency:  2.491200411267457e+09

...

 	 Residual:   2.54e-11

...


eigenvalue:

...

  3.545447315320415e+03

...

 	 Frequency:  2.841033486606133e+09

...

 	 Residual:   1.25e-09

Note that the residual of the eigenpair should be reasonably small to be a good solution.