Running four serial jobs at the same time on the same computer

JuPa · August 7, 2014, 08:14

Hi,

I have a single phase case with around 30K cells. It won't benefit from parralel speed up. I also have a four core machine laying around at work.

Is it okay to run this case four times (with four different boundary conditions) all in parallel with each other?

I.e. Are there any issues with running 4 serial jobs at the same time?

Will it be dog slow?

The machine is an i7 quad core machine with 8 gb RAM.

Thanks.

Lance · August 7, 2014, 08:37

If running a parallell job on all four cores works I dont see why there should be any problems with running four serial at the same time. Why not try and see for yourself? Maybe would run three jobs and leave the last core for the OS?

ghorrocks · August 7, 2014, 09:44

It is always quicker to run 4 serial jobs in parallel compared to a series of 4-way parallel jobs in series. And in your case as the size is quite small this will be especially the case.

The only disadvantage is you need 4 solver licenses to do this, the parallel option requires a single solver license and some parallel options. The 4 solver licenses are much more expensive.

JuPa · August 7, 2014, 10:29

Thanks for the replies. I'll get cracking.

JuPa · August 11, 2014, 06:36

I've run into a few problems. I started four transient serial jobs at the same time. All four jobs are the same except from the turbulence model used. The models I am using are:
- RNG K Eps
- SST
- BSL RSM
- BSL EARSM

The SST job failed citing:

Code:

 Fatal bounds error detected
  ---------------------------
  Variable: Absolute Pressure
  Locale  : Default Domain

 +--------------------------------------------------------------------+
 |                    Writing crash recovery file                     |
 +--------------------------------------------------------------------+
                             
  Fatal bounds error detected
  ---------------------------
  Variable: Absolute Pressure
  Locale  : Default Domain

 +--------------------------------------------------------------------+
 |                An error has occurred in cfx5solve:                 |
 |                                                                    |
 | The ANSYS CFX solver exited with return code 1.   No results file  |
 | has been created.                                                  |
 +--------------------------------------------------------------------+

End of solution stage.

 +--------------------------------------------------------------------+
 | The following transient and backup files written by the ANSYS CFX  |
 | solver have been saved in the directory                            |
 | /usr/HomogeneousConvection/Trans-Conv-SST_001: |
 |                                                                    |
 | 355_full.trn, 2527_full.bak                                        |
 +--------------------------------------------------------------------+


 +--------------------------------------------------------------------+
 |                An error has occurred in cfx5solve:                 |
 |                                                                    |
 | Error reported by IO module: setfilebusyflag: more than one user   |
 | is reading                                                         |
 | /usr/HomogeneousConvection/Trans-Conv-SST_001- |
 | .dir/crash or the wrong lock file                                  |
 | /usr/HomogeneousConvection/Trans-Conv-SST_001- |
 | .dir/crash.lck exists                                              |
 +--------------------------------------------------------------------+


 +--------------------------------------------------------------------+
 |                An error has occurred in cfx5solve:                 |
 |                                                                    |
 | Error reported by IO module: iocnt: setfilebusyflag failed when    |
 | writing a data set header                                          |
 +--------------------------------------------------------------------+


 +--------------------------------------------------------------------+
 |                An error has occurred in cfx5solve:                 |
 |                                                                    |
 | Error from IO module initializing block write to G/MON EVERY 2529: |
 |                                                                    |
 |  "iocnt: setfilebusyflag failed when writing a data set header"    |
 |                                                                    |
 | unknown(?): Successful.                                            |
 +--------------------------------------------------------------------+


 +--------------------------------------------------------------------+
 |                An error has occurred in cfx5solve:                 |
 |                                                                    |
 | Error from IO module while writing G/MON:EVERY:2529:               |
 |                                                                    |
 |  "iocnt: setfilebusyflag failed when writing a data set header"    |
 |                                                                    |
 | unknown(?): Successful.                                            |
 +--------------------------------------------------------------------+


 +--------------------------------------------------------------------+
 |                              Warning!                              |
 |                                                                    |
 | The ANSYS CFX Solver has written a crash recovery file.  This file |
 | has been saved as                                                  |
 | /usr/HomogeneousConvection/Trans-Conv-SST_001- |
 | .res.err and may be an aid to diagnosing the problem or restarting |
 | the run.  More details should be available in the solver output    |
 | section of the output file.  Note that a lock file was left for    |
 | the crash file when the solver exited, and so it is probably       |
 | incomplete.                                                        |
 +--------------------------------------------------------------------+


 +--------------------------------------------------------------------+
 | The following user files have been saved in the directory          |
 | /usr/HomogeneousConvection/Trans-Conv-SST_001: |
 |                                                                    |
 | mon                                                                |
 +--------------------------------------------------------------------+


This run of the ANSYS CFX Solver has finished.

The RNG job failed citing:

Code:

                             
  Fatal bounds error detected
  ---------------------------
  Variable: Absolute Pressure
  Locale  : Default Domain

 +--------------------------------------------------------------------+
 |                    Writing crash recovery file                     |
 +--------------------------------------------------------------------+
                             
  Fatal bounds error detected
  ---------------------------
  Variable: Absolute Pressure
  Locale  : Default Domain

 +--------------------------------------------------------------------+
 |                An error has occurred in cfx5solve:                 |
 |                                                                    |
 | The ANSYS CFX solver exited with return code 1.   No results file  |
 | has been created.                                                  |
 +--------------------------------------------------------------------+

End of solution stage.

 +--------------------------------------------------------------------+
 | The following transient and backup files written by the ANSYS CFX  |
 | solver have been saved in the directory                            |
 | /usr/HomogeneousConvection/Trans-Conv-RNG_001: |
 |                                                                    |
 | 355_full.trn, 2172_full.bak                                        |
 +--------------------------------------------------------------------+


 +--------------------------------------------------------------------+
 |                An error has occurred in cfx5solve:                 |
 |                                                                    |
 | Error reported by IO module: setfilebusyflag: more than one user   |
 | is reading                                                         |
 | /usr/HomogeneousConvection/Trans-Conv-RNG_001- |
 | .dir/crash or the wrong lock file                                  |
 | /usr/HomogeneousConvection/Trans-Conv-RNG_001- |
 | .dir/crash.lck exists                                              |
 +--------------------------------------------------------------------+


 +--------------------------------------------------------------------+
 |                An error has occurred in cfx5solve:                 |
 |                                                                    |
 | Error reported by IO module: iocnt: setfilebusyflag failed when    |
 | writing a data set header                                          |
 +--------------------------------------------------------------------+


 +--------------------------------------------------------------------+
 |                An error has occurred in cfx5solve:                 |
 |                                                                    |
 | Error from IO module initializing block write to G/MON EVERY 2184: |
 |                                                                    |
 |  "iocnt: setfilebusyflag failed when writing a data set header"    |
 |                                                                    |
 | unknown(?): Successful.                                            |
 +--------------------------------------------------------------------+


 +--------------------------------------------------------------------+
 |                An error has occurred in cfx5solve:                 |
 |                                                                    |
 | Error from IO module while writing G/MON:EVERY:2184:               |
 |                                                                    |
 |  "iocnt: setfilebusyflag failed when writing a data set header"    |
 |                                                                    |
 | unknown(?): Successful.                                            |
 +--------------------------------------------------------------------+


 +--------------------------------------------------------------------+
 |                              Warning!                              |
 |                                                                    |
 | The ANSYS CFX Solver has written a crash recovery file.  This file |
 | has been saved as                                                  |
 | /usr/HomogeneousConvection/Trans-Conv-RNG_001- |
 | .res.err and may be an aid to diagnosing the problem or restarting |
 | the run.  More details should be available in the solver output    |
 | section of the output file.  Note that a lock file was left for    |
 | the crash file when the solver exited, and so it is probably       |
 | incomplete.                                                        |
 +--------------------------------------------------------------------+


 +--------------------------------------------------------------------+
 | The following user files have been saved in the directory          |
 | /usr/HomogeneousConvection/Trans-Conv-RNG_001: |
 |                                                                    |
 | mon                                                                |
 +--------------------------------------------------------------------+


This run of the ANSYS CFX Solver has finished.

The BSL EARSM and BSL RSM jobs are ticking along nicely. My initial guess is that the two jobs tried to write to the hard drive at the same time and failed. Does anyone have any opinions?

ghorrocks · August 11, 2014, 07:40

No - you have a bounds error. This means the pressure went outside of allowable limits - presumably your pressure went below absolute zero.

This means those simulations were numerically unstable.

JuPa · August 11, 2014, 07:50

Quote:

Originally Posted by ghorrocks

No - you have a bounds error. This means the pressure went outside of allowable limits - presumably your pressure went below absolute zero.

This means those simulations were numerically unstable.

Bollocks.

I'll dig deeper into the setup/boundary conditions to find out what the issue is. I'm using variable properties IAPWS IF97 water. It's given me issues before. It might be that. But I'll have a look. Thanks.

ghorrocks · August 11, 2014, 07:56

You are using IAPWS? Then you definitely have a bounds error. These complex material property models are always hard to converge.

JuPa · August 11, 2014, 08:02

Quote:

Originally Posted by ghorrocks

You are using IAPWS? Then you definitely have a bounds error. These complex material property models are always hard to converge.

Aye. I've had many a late night trying to get variable props water to converge properly.

August 7, 2014, 08:14	Running four serial jobs at the same time on the same computer	#1
JuPa Senior Member Mr CFD Join Date: Jun 2012 Location: Britain Posts: 361 Rep Power: 15	Hi, I have a single phase case with around 30K cells. It won't benefit from parralel speed up. I also have a four core machine laying around at work. Is it okay to run this case four times (with four different boundary conditions) all in parallel with each other? I.e. Are there any issues with running 4 serial jobs at the same time? Will it be dog slow? The machine is an i7 quad core machine with 8 gb RAM. Thanks.

August 11, 2014, 07:40		#6
ghorrocks Super Moderator Glenn Horrocks Join Date: Mar 2009 Location: Sydney, Australia Posts: 17,854 Rep Power: 144	No - you have a bounds error. This means the pressure went outside of allowable limits - presumably your pressure went below absolute zero. This means those simulations were numerically unstable. JuPa likes this.

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
License trouble: ANSLIC_ADMIN, FLEXlm not running, but two lmgrd processes running!	dancoz87	ANSYS	17	November 14, 2016 02:10
Running dieselFoam in parallel.	Palminchi	OpenFOAM	0	February 17, 2010 05:00
Icem tetra - Running in parrallel?	Joe	CFX	5	September 9, 2006 13:37
star is not running the simulation in windows	Arnab	Siemens	1	August 2, 2004 03:40
Running 2 CFD jobs on one PC	steve podleski	Main CFD Forum	17	February 16, 2000 15:40

August 7, 2014, 08:37		#2
Lance Senior Member Lance Join Date: Mar 2009 Posts: 669 Rep Power: 22	If running a parallell job on all four cores works I dont see why there should be any problems with running four serial at the same time. Why not try and see for yourself? Maybe would run three jobs and leave the last core for the OS?

August 7, 2014, 09:44		#3
ghorrocks Super Moderator Glenn Horrocks Join Date: Mar 2009 Location: Sydney, Australia Posts: 17,854 Rep Power: 144	It is always quicker to run 4 serial jobs in parallel compared to a series of 4-way parallel jobs in series. And in your case as the size is quite small this will be especially the case. The only disadvantage is you need 4 solver licenses to do this, the parallel option requires a single solver license and some parallel options. The 4 solver licenses are much more expensive.

August 7, 2014, 10:29		#4
JuPa Senior Member Mr CFD Join Date: Jun 2012 Location: Britain Posts: 361 Rep Power: 15	Thanks for the replies. I'll get cracking.

August 11, 2014, 07:56		#8
ghorrocks Super Moderator Glenn Horrocks Join Date: Mar 2009 Location: Sydney, Australia Posts: 17,854 Rep Power: 144	You are using IAPWS? Then you definitely have a bounds error. These complex material property models are always hard to converge.