Slow NX performance

Pioneer
Pioneer

Friends,

 

I am fortunate to have access to both ANSYS and FEMAP with Nx Nastran at my day job. We have been requested to cut down on the number of ANSYS processing licenses as a cost-cutting initiative. Thus I have begun experimenting with models and core combinations to determine the optimum number of cores to provide timely analyses while minimizing waste. We are currently using older HP workstations Running Windows 7, these workstations have 128 Gb of RAM, 1TB SSD NVME and dual Xeon processors E5-2643V3 or E5-2670 depending on the vintage.I began this process by setting up a simple linear static model with 225000 elements and 948321 nodes and solving the model with default memory settings with 1, 2, 4, 6, 8 and 12 CPUs. I was very pleased with the results the ANSYS model solved in less than 1 min for 6, 8 and 12 CPUs. The difference between 6, 8 and 12 CPUs was minor and I plan on testing large models in the near future. If anyone is interested I can see request permission to share the results. 

 

So now for my question, after experimenting with ANSYS I decided to see how long it would take FEMAP 11.1.2 /NXNASTRAN 9.1 to solve the exact same model. I created an identical simple linear static model with 225000 elements and 948321 nodes and began solving the model with default memory settings with 1, 2, 4, 6, 8 and 12 CPUs. I used the Elemental Iterative Solver and I'm seeing solve times in excess of 6 minutes. I assume I am using the incorrect solver or do not have my memory allocated properly. I have my RCF file set to use 45% of my physical memory, the buffer size is set for 65537, my scratch drive is my NVME SSD, and I'm requesting 12 processors. Is there any way for ANSYS solver to be 5.9 times faster than the NX solver? Please tell me something isn't set up properly? 

 

Thanks

John 

 

23 REPLIES 23

Re: Slow NX performance

Gears Phenom Gears Phenom
Gears Phenom

John

 

Have you set your solver to the ILP-64bit, otherwise NX Nastran will only ever use 8Gb of RAM.

 

IPL-64.JPG

Re: Slow NX performance

Siemens Phenom Siemens Phenom
Siemens Phenom

Why are you benchmarking with NX Nastran 9.1? That is 8 releases / 4 years old?

 

Can you post the .f04/.f06 and .log files from one of the Nastran runs?

 

 

 

 

 

 

Re: Slow NX performance

Pioneer
Pioneer

 

 

I was running the NX 9.1 as this is what was already loaded on the computer, I understand it is 4 years old. I have since my last posting downloaded FEMAP 11.4.2 with NX 11.0.2 and began the same testing regime the results are quite similar. I have attached the F04 file as requested, my RCF file, and a graph showing differences between the various products tested, please note that both NX Nastran 9.1 and 11.0.2 were solved with the Elemental Iterative Solver. Please look at the files and images, I'm hoping that I do not have something setup properly. 

 

Thanks

 

John

 

ILP 64.PNGILP-64

RCF.PNGRCF File NX 11.0.2


Speeds.PNGSolver Performance Comparison

Re: Slow NX performance

Phenom
Phenom

Is there a way you could post your model to have others run on difference version of NX Nastran (and possibly other solvers)? There shouldn't be that much difference between ANSYS and NX Nastran in simple linear statics...

 

Is it the same type of elements (i.e. no reduced integration trick on the ANSYS side)?  Is Nastran running through GEOMCHECK while ANSYS isn't? AUTOSPC? AUTOMPC?

Re: Slow NX performance

Pioneer
Pioneer

TentechLLC,

 

Thanks for taking a look at these files. They are simple files and I do not believe there is a discrepancy between how the ANSYS WB and NxNastran files are setup in each respective program. I would appreciate your expert opinion, and if I do have something setup incorrectly please let me know.

 

Thank you for your help.

 

JP 

Re: Slow NX performance

Phenom
Phenom

JP

Could you post the nx.nastran input file? One might be able to run it to see how it compares with you run

Production: NX.CAE 10.0.2.6
Development: VB.NET (amateur level !)

Re: Slow NX performance

Solution Partner Phenom Solution Partner Phenom
Solution Partner Phenom

Dear John,

I have just solved the posted model file using FEMAP V11.4.2 and NX NASTRAN V11.0.2 with the elemental iterative solver & IPL64, of course, and the model solved in 1min 30 sec in my "old" laptop with 32 GB RAM and i7 quadcore i7-3840QM CPU @ 2.80GH, nothing special settings, using PARALLEL=4, attached *.LOG, *.F04, etc.. files.

In any modern processor with plenty of cores & processor speed the solution time should be less than one minute for sure!!.

 *** PARALLEL PROCESSES CPU INFORMATION (SECONDS) ***

  MODULE     PROCESS-1     PROCESS-2     PROCESS-3     PROCESS-4     PROCESS-5     PROCESS-6     PROCESS-7     PROCESS-8
  --------   -----------   -----------   -----------   -----------   -----------   -----------   -----------   -----------
  P-CASI          8.89          8.89          8.89          8.89
  ========   ===========   ===========   ===========   ===========   ===========   ===========   ===========   ===========
  TOTAL           8.89          8.89          8.89          8.89

 *** TOTAL MEMORY AND DISK USAGE STATISTICS ***

 +---------- SPARSE SOLUTION MODULES -----------+         +------------- MAXIMUM DISK USAGE -------------+
      HIWATER               SUB_DMAP        DMAP               HIWATER               SUB_DMAP        DMAP
      (WORDS)   DAY_TIME      NAME         MODULE                 (MB)   DAY_TIME      NAME         MODULE
    836034704   09:46:51    SOL2      368  SOLVIT              666.375   09:47:20    SOL2      600  EXIT    


 *** DATABASE USAGE STATISTICS ***
 
 +------------------ LOGICAL DBSETS ------------------+   +------------------------- DBSET FILES -------------------------+
  DBSET      ALLOCATED   BLOCKSIZE        USED    USED     FILE       ALLOCATED     HIWATER       HIWATER  I/O TRANSFERRED
              (BLOCKS)     (WORDS)    (BLOCKS)       %                 (BLOCKS)    (BLOCKS)          (MB)             (GB)
 
  MASTER          5000       65536          48    0.96     MASTER          5000          48        24.000            0.503
  DBALL        2000000       65536        1243    0.06     DBALL        2000000        1243       621.500            0.613
  OBJSCR          5000        8192         318    6.36     OBJSCR          5000         318        19.875            0.025
  SCRATCH      4005878       65536        1250    0.03    (MEMFILE         5878        2308      1154.000            0.000)
                                                           SCRATCH      2000000           1         0.500            0.000
                                                           SCR300       2000000           1         0.500            0.000
                                                                                                            ==============
                                                                                                     TOTAL:          1.142
 

SPEED-TEST-ITERATIVE-SOLVER.png

Best regards,
Blas.

Blas Molero Hidalgo, Ingeniero Industrial, Director
IBERISA • 48004 BILBAO (SPAIN)
WEB: http://www.iberisa.com
Blog Femap-NX Nastran: http://iberisa.wordpress.com/

Re: Slow NX performance

Phenom
Phenom

3minutes on my machine Man Sad

using NX.nastran 10. run .dat file as provided by Blas

 MODULE     PROCESS-1     PROCESS-2     PROCESS-3     PROCESS-4     PROCESS-5     PROCESS-6     PROCESS-7     PROCESS-8
  --------   -----------   -----------   -----------   -----------   -----------   -----------   -----------   -----------
  P-CASI          9.00          9.00          9.00          9.00          9.00          9.00          9.00          9.00
  ========   ===========   ===========   ===========   ===========   ===========   ===========   ===========   ===========
  TOTAL           9.00          9.00          9.00          9.00          9.00          9.00          9.00          9.00

  MODULE     PROCESS- 9    PROCESS-10    PROCESS-11    PROCESS-12    PROCESS-13    PROCESS-14    PROCESS-15    PROCESS-16
  --------   -----------   -----------   -----------   -----------   -----------   -----------   -----------   -----------
  P-CASI          9.00          9.00          9.00          9.00          9.00          9.00          9.00          9.00
  ========   ===========   ===========   ===========   ===========   ===========   ===========   ===========   ===========
  TOTAL           9.00          9.00          9.00          9.00          9.00          9.00          9.00          9.00

  MODULE     PROCESS-17    PROCESS-18    PROCESS-19    PROCESS-20    PROCESS-21    PROCESS-22    PROCESS-23    PROCESS-24
  --------   -----------   -----------   -----------   -----------   -----------   -----------   -----------   -----------
  P-CASI          9.00          9.00          9.00          9.00          9.00          9.00          9.00          9.00
  ========   ===========   ===========   ===========   ===========   ===========   ===========   ===========   ===========
  TOTAL           9.00          9.00          9.00          9.00          9.00          9.00          9.00          9.00

  MODULE     PROCESS-25    PROCESS-26    PROCESS-27    PROCESS-28    PROCESS-29    PROCESS-30    PROCESS-31    PROCESS-32
  --------   -----------   -----------   -----------   -----------   -----------   -----------   -----------   -----------
  P-CASI          9.00          9.00          9.00          9.00          9.00          9.00          9.00          9.00
  ========   ===========   ===========   ===========   ===========   ===========   ===========   ===========   ===========
  TOTAL           9.00          9.00          9.00          9.00          9.00          9.00          9.00          9.00


 *** TOTAL MEMORY AND DISK USAGE STATISTICS ***

 +---------- SPARSE SOLUTION MODULES -----------+         +------------- MAXIMUM DISK USAGE -------------+
      HIWATER               SUB_DMAP        DMAP               HIWATER               SUB_DMAP        DMAP
      (WORDS)   DAY_TIME      NAME         MODULE                 (MB)   DAY_TIME      NAME         MODULE
  12427921067   09:33:12    SOL2      368  SOLVIT              663.562   09:35:01    SOL2      604  EXIT    


 *** DATABASE USAGE STATISTICS ***
 
 +------------------ LOGICAL DBSETS ------------------+   +------------------------- DBSET FILES -------------------------+
  DBSET      ALLOCATED   BLOCKSIZE        USED    USED     FILE       ALLOCATED     HIWATER       HIWATER  I/O TRANSFERRED
              (BLOCKS)     (WORDS)    (BLOCKS)       %                 (BLOCKS)    (BLOCKS)          (MB)             (GB)
 
  MASTER          5000       65536          46    0.92     MASTER          5000          46        23.000            0.443
  DBALL        2000000       65536        1243    0.06     DBALL        2000000        1243       621.500            0.613
  OBJSCR          5000        8192         289    5.78     OBJSCR          5000         289        18.062            0.022
  SCRATCH      4146713       65536        1250    0.03    (MEMFILE       146713        2308      1154.000            0.000)
                                                           SCRATCH      2000000           1         0.500            0.000
                                                           SCR300       2000000           1         0.500            0.000
                                                                                                            ==============
                                                                                                     TOTAL:          1.079
Production: NX.CAE 10.0.2.6
Development: VB.NET (amateur level !)

Re: Slow NX performance

Solution Partner Phenom Solution Partner Phenom
Solution Partner Phenom

Dear Select,

Reduce the number of proccess, try PARALLEL=8 and run the analysis again, make sure that intel HyperThreading is not active in your machine´s BIOS, from my tests with number of cores in the past I learned that too much is less, and depending the solver (not the same SOL101 or SOL106) the scalability is not the same.

Best regards,
Blas.


@selex_ct wrote:

3minutes on my machine Man Sad

using NX.nastran 10. run .dat file as provided by Blas


 

Blas Molero Hidalgo, Ingeniero Industrial, Director
IBERISA • 48004 BILBAO (SPAIN)
WEB: http://www.iberisa.com
Blog Femap-NX Nastran: http://iberisa.wordpress.com/