This section outlines the additional information that is written to the CFX-Solver Output file for CFX-Solver jobs submitted in parallel.
If the partitioning step is run, partitioning information pertaining to the current job is displayed. This includes how the mesh is divided, and CPU requirements for the partitioning process.
The section for partitioning information looks similar to the following:
+--------------------------------------------------------------------+
| Vertex Based Partitioning |
+--------------------------------------------------------------------+
Partitioning of domain: Default Domain
- Partitioning tool: MeTiS multilevel k-way algorithm
- Number of partitions: 8
- Number of graph-nodes: 5232026
- Number of graph-edges: 56867380
+--------------------------------------------------------------------+
| Partitioning Information |
+--------------------------------------------------------------------+
Partitioning information for domain: Default Domain
+------------------+------------------------+-----------------+
| Elements | Vertices | Faces |
+------+------------------+------------------------+-----------------+
| Part | Number % | Number % %Ovlp | Number % |
+------+------------------+------------------------+-----------------+
| Full | 20584862 | 5232026 | 235752 |
+------+------------------+------------------------+-----------------+
| 1 | 2592908 12.5 | 675743 12.6 2.9 | 27947 11.7 |
| 2 | 2581847 12.4 | 672141 12.5 2.5 | 28030 11.8 |
| 3 | 2752840 13.2 | 666862 12.4 1.3 | 31200 13.1 |
| 4 | 2600679 12.5 | 663980 12.4 2.5 | 27997 11.8 |
| 5 | 2829665 13.6 | 678465 12.7 3.3 | 21784 9.2 |
| 6 | 2027720 9.7 | 666910 12.5 2.6 | 43374 18.2 |
| 7 | 2696854 13.0 | 670512 12.5 2.3 | 25832 10.9 |
| 8 | 2721802 13.1 | 661733 12.4 1.2 | 31858 13.4 |
+------+------------------+------------------------+-----------------+
| Min | 2027720 9.7 | 661733 12.4 1.2 | 21784 9.2 |
|(part)| ( 6)| ( 8 8)| ( 5)|
+------+------------------+------------------------+-----------------+
| Max | 2829665 13.6 | 678465 12.7 3.3 | 43374 18.2 |
|(part)| ( 5)| ( 5 5)| ( 6)|
+------+------------------+------------------------+-----------------+
| Ave | 2600539 12.5 | 669543 12.5 2.3 | 29753 12.5 |
+------+------------------+------------------------+-----------------+
| Sum | 20804315 100.0 | 5356346 100.0 2.3 | 238022 100.0 |
+------+------------------+------------------------+-----------------+
+--------------------------------------------------------------------+
| Partitioning CPU-Time Requirements |
+--------------------------------------------------------------------+
- Preparations 1.643E+01 seconds
- Low-level mesh partitioning 5.711E+00 seconds
- Global partitioning information 5.219E-01 seconds
- Element and face partitioning information 1.142E+00 seconds
- Vertex partitioning information 3.639E-01 seconds
- Partitioning information compression 5.499E-02 seconds
- Summed CPU-time for mesh partitioning 2.471E+01 secondsThis section describes the job characteristics in terms of the parallel run mode and the time and date of the start of the run.
The section for job information for a parallel run looks similar to the following:
+--------------------------------------------------------------------+ | Job Information at Start of Run | +--------------------------------------------------------------------+ Run mode: parallel run (MPI) Job started: Tue Dec 6 11:23:46 2016
Note: Occasionally, follower processes may continue to run after the leader process
has terminated. More detailed job information, including the host names,
process IDs, start time, finish time and CPU time for each partition, is
written to a temporary file named
"pids", which exists during the
lifetime of the run. If the run terminates successfully, this information is
written to the results file and may be read using the command
"cfx5dfile -read-parinfo".
Note: The above format is the default output format. A more verbose format, as used in earlier releases of Ansys CFX, is available. For details, see Command-Line Options and Keywords for cfx5solve.
The initial memory allocated on each host is reported near the start of the run, similar to the following:
+--------------------------------------------------------------------+ | Host Memory Information (Mbytes): Solver | +--------------------------------------------------------------------+ | Host | Npart | System | Allocated | % | +------------------------+-------+-------------+-------------+-------+ | fastcomputer1 | 32 | 32307.24 | 9280.91 | 28.73 | | fastcomputer2 | 32 | 32307.24 | 9195.09 | 28.46 | | fastcomputer3 | 32 | 32307.24 | 9713.21 | 30.07 | | fastcomputer4 | 32 | 32307.24 | 9409.84 | 29.13 | +------------------------+-------+-------------+-------------+-------+ | Total | 128 | 129228.97 | 37599.05 | 29.09 | +------------------------+-------+-------------+-------------+-------+
Similarly, the peak memory usage for each host over the lifetime of the run is reported near the end of the run. For example:
+--------------------------------------------------------------------+ | Host Memory Information (Mbytes): Solver | +--------------------------------------------------------------------+ | Host | Npart | System | Peak | % | +------------------------+-------+-------------+-------------+-------+ | fastcomputer1 | 32 | 32307.24 | 10673.05 | 33.04 | | fastcomputer2 | 32 | 32307.24 | 10574.35 | 32.73 | | fastcomputer3 | 32 | 32307.24 | 11170.19 | 34.57 | | fastcomputer4 | 32 | 32307.24 | 10821.32 | 33.50 | +------------------------+-------+-------------+-------------+-------+ | Total | 128 | 129228.97 | 43238.91 | 33.46 | +------------------------+-------+-------------+-------------+-------+
The default output format produces a section for memory usage information that looks similar to the following:
+--------------------------------------------------------------------+
| Initial Memory Allocation (Actual usage may vary) |
+--------------------------------------------------------------------+
Allocated storage in: Mwords
Mbytes
Partition | Real | Integer | Character | Logical | Double
----------+------------+------------+-----------+----------+----------
Minimum | 8.88 | 1.74 | 3.70 | 0.12 | 0.09
( 4) | 33.86 | 6.65 | 3.53 | 0.11 | 0.69
----------+------------+------------+-----------+----------+----------
Maximum | 8.90 | 1.75 | 3.70 | 0.12 | 0.09
( 1) | 33.96 | 6.68 | 3.53 | 0.11 | 0.69
----------+------------+------------+-----------+----------+----------
Average | 8.89 | 1.75 | 3.70 | 0.12 | 0.09
| 33.89 | 6.66 | 3.53 | 0.11 | 0.69
----------+------------+------------+-----------+----------+----------
Total | 35.54 | 6.99 | 14.81 | 0.48 | 0.36
| 135.58 | 26.65 | 14.12 | 0.46 | 2.76
----------+------------+------------+-----------+----------+----------The section for completed job information looks similar to the following:
+--------------------------------------------------------------------+
| Job Information at End of Run |
+--------------------------------------------------------------------+
Job finished: Tue Dec 3 11:23:56 2013
Total wall clock time: 9.630E+00 seconds
or: ( 0: 0: 0: 9.630 )
( Days: Hours: Minutes: Seconds )
--> Final synchronization point reached by all partitions.
End of solution stage.
+--------------------------------------------------------------------+
| The results from this run of the CFX-Solver have been written |
| to z:\temp\BluntBody_001.res |
+--------------------------------------------------------------------+Note: The above format is the default output format. A more verbose format, as used in earlier releases of Ansys CFX is available. For details, see Command-Line Options and Keywords for cfx5solve.