[ramune]

Thank you

The problem was the settings for graphic card.
$ WM_PROJECT_DIR / wmake / rules / linux64Nvcc
C and c ++
CC = nvcc-Xptxas-dlcm = cg-m64-gencode = arch = compute_70, code = sm_70 -arch = sm_70
If you change to, it will work without error.

However, it is slower than running on my own PC, so I will try to solve this problem.
I think that the content is different from this thread, so I will try my best.

[agustinvo]

Hi
I have downloaded RapidCFD, and also CUDA 9.1, which is the one my laptop supports. Yes, maybe it is not the optimal place to run large cases, but I can to see how RapidCFD is working and test its adequacy in the future for me.


Once I have installed CUDA, RapidCFD, and I start the installation, I get the error saying that my compiler GCC is newer than the one supported by CUDA9.1 (gcc 6). How can I get through this problem? did you experience the same before?


I will post later the log file.

[edaymo]

agustinvo -
On the gcc compile error... there is a table of compatible gcc versions for compiling with CUDA. E.g, for CUDA 9.1, the compatible gcc versions are shown at (https://docs.nvidia.com/cuda/archive...nux/index.html). For the latest version of CUDA (now 10.1 update1), the table of compatible gcc versions is here, (https://docs.nvidia.com/cuda/archive...nux/index.html). Hopefully you can match your gcc compiler to the CUDA version + OS you are using.


ramune - Congratulations on solving your compile problem! On your speed issue, as has been discussed in these forums, you need a really large grid to see speed improvements with RapidCFD. Inspired by the post (Comparison of OpenFOAM on i7, Xeon@32 cores, Xeon Phi Knights Landing, Tesla K20m) post, I recreated the damBreak case and placed it here, (https://github.com/TonkomoLLC/RapidCFD-Tests). Hopefully this case will help you with speed testing.

[agustinvo]

Thank you edaymo for your answer. I followed the steps given here:


https://stackoverflow.com/questions/...my-gcc-version

For CUDA 9:

then you make the link so CUDA takes the good compiler gcc 6

Now I'm facing the error

terminate called after throwing an instance of 'thrust::system::system_error'
what(): parallel_for failed: no kernel image is available for execution on the device

I saw it already in this thread, I'll check now

[edaymo]

Hi, agustinvo -

Hmm, I don't think I personally encountered the error "what(): parallel_for failed: no kernel image is available for execution on the device"

As you mentioned, there are some posts on this error on the internet. One that I read suggests that you might get this error if your GPU is of compute capability < 3 (I think RapidCFD requires at least compute capability > 3).

Anyhow, I hope this lead is helpful.

I hope you find the solution.

Best regards,
Eric

[ramune]

Hi,Eric

Thanks for your reply.
I want to do this benchmark , but I think RapidCFD doesn't include MeshTools. Of course, It is possible to install OpenFOAM and use MeshTools. I don't know How to use MeshTools in RapidCFD.

For now, I use MeshTools in OpenFOAM before using RapidCFD.

[edaymo]

Hi, Ramune,

Yes, you are correct. RapidCFD does not have blockMesh, etc. So you must use utilities (paraFoam, setFields, blockMesh, and so on) from standard CPU OpenFOAM (eg., 2.3.x).

Best regards,

Eric

[agustinvo]

Originally Posted by edaymo

Hi, agustinvo -

Hmm, I don't think I personally encountered the error "what(): parallel_for failed: no kernel image is available for execution on the device"

As you mentioned, there are some posts on this error on the internet. One that I read suggests that you might get this error if your GPU is of compute capability < 3 (I think RapidCFD requires at least compute capability > 3).

Anyhow, I hope this lead is helpful.

I hope you find the solution.

Best regards,
Eric

Hi Eric


As you say, the compute capability could be the problem: my humble GeForce 710M has a compute capability of 2.1... But checking the Issues in the GitHub, I mhight have found a way to use it:


https://github.com/Atizar/RapidCFD-dev/issues/59


which is to modify the flags of the nvcc. I will check it later on during the day.


And yes, this graphic card it could not be the best (already 6 years), but I want to play a bit and get familiar to this fork.

[edaymo]

Hi, agustinvo

I hope that issue helps solve the problem. I read about the requirement of CC > 3 at:

https://github.com/Atizar/RapidCFD-dev/issues/29

Personally, I have not tested CC < 3, so I cannot comment from personal experience.

If you find a solution please let us know.

Best regards,

Eric

[ramune]

Hi, Eric

I used blockMesh and setFields with OpenFOAM-v1812.
After that, I used interFoam. Error was occured.

Create time

Create mesh for time = 0

terminate called after throwing an instance of 'thrust::system::system_error'
what(): __copy:: H->D: failed: invalid argument

Thank you.

[edaymo]

Hello, Ramune,

Please also see: https://github.com/Atizar/RapidCFD-dev/issues/56

I have seen this error when there is insufficient memory on the GPU for the case you are trying to run.

To confirm, re-create the grid with fewer cells. Hopefully RapidCFD will run with a case that is appropriately sized for your GPU memory.

Best regards,

Eric

[agustinvo]

Originally Posted by edaymo

Hi, agustinvo

I hope that issue helps solve the problem. I read about the requirement of CC > 3 at:

https://github.com/Atizar/RapidCFD-dev/issues/29

Personally, I have not tested CC < 3, so I cannot comment from personal experience.

If you find a solution please let us know.

Best regards,

Eric

Hi Eric,
unfortunately, I tried to install ir using CC=2.1, but it didn't pass from the Allwmake command. It seems it is not possible to use this card for RapidCFD.


Anyway, thanks for your comments! I hope one day I will have a good enough card to test it.

[lb0510]

This is most likely a very basic issue but I did not find any other instructions on how to fix it. Maybe this issue will result in the instructions in the README file being updated.
So I cloned the Repo and tried installing it with ./Allwmake.

wmakeCheckPwd error: Current directory is not /opt/openfoam6
Error: Current directory is not $WM_PROJECT_DIR
The environment variables are inconsistent with the installation.
Check the OpenFOAM entries in your dot-files and source them.

The problem is, that my current directory indeed was /opt/openfoam6. I even tried copying RapidCFD to /opt/ and to set $FOAM_INSTALL to the RapidCFD directory location.
Any help is appreciated!

[mm66]

Dear Foamers (RapidCFDers?!),


I eventually managed to compile RapidCFD and successfully ran a simulation using icoFoam. My next step is to run a simulation with chtMultiRegionSimpleFoam but it seems missing as I get:

Command 'chtMultiRegionSimpleFoam' not found, but can be installed with:

sudo apt install openfoam

Please help.


Regards,
MJ

[edaymo]

Hi, MJ,

If you go to `$WM_PROJECT_DIR/applications/solvers/heatTransfer/chtMultiRegionFoam`, please try`wmake` and then repeat `wmake` in the `chtMultiRegionSimpleFoam` directory.


I never tried chtMultiRegionFoam with RapidCFD, but I noticed on my computer that it was not compiled. I was able to compile the transient and steady state CHT solvers by just going to the appropriate source directories and typing `wmake.`

Hope you are successful in getting chtMultiRegionSimpleFoam to work.

Best regards,

Eric

[mm66]

Originally Posted by edaymo

Hi, MJ,

If you go to `$WM_PROJECT_DIR/applications/solvers/heatTransfer/chtMultiRegionFoam`, please try`wmake` and then repeat `wmake` in the `chtMultiRegionSimpleFoam` directory.


I never tried chtMultiRegionFoam with RapidCFD, but I noticed on my computer that it was not compiled. I was able to compile the transient and steady state CHT solvers by just going to the appropriate source directories and typing `wmake.`

Hope you are successful in getting chtMultiRegionSimpleFoam to work.

Best regards,

Eric

Hi Eric,


Your solution worked and I really really appreciate it


But (hate to say "but"), now I am facing another problem:

-> FOAM FATAL IO ERROR: 
Unknown patchField type compressible::turbulentTemperatureCoupledBaffleMixed for patch type mappedWall

Turns out RapidCFD does not support this boundary condition Any suggestions instead?



Thanks a lot.


Regards,
MJ

[edaymo]

Hi, MJ,

Looks like in:

https://github.com/Atizar/RapidCFD-d...del/Make/files


the BC you want is commented out, so it's not compiled.

I tried uncommenting the BC you want, and compilation failed.

Sorry, but it is probably some work to get this to work. I have no idea if it is a little or a lot of effort.

Best regards,

Eric

[mm66]

Originally Posted by edaymo

Hi, MJ,

Looks like in:

https://github.com/Atizar/RapidCFD-d...del/Make/files


the BC you want is commented out, so it's not compiled.

I tried uncommenting the BC you want, and compilation failed.

Sorry, but it is probably some work to get this to work. I have no idea if it is a little or a lot of effort.

Best regards,

Eric

Hi Eric,


I really appreciate your time. This helps me know where to begin. Hope I can get that beast running. Thank you very much.


Cheers,
MJ

[mm66]

Originally Posted by edaymo

Hi, MJ,

Looks like in:

https://github.com/Atizar/RapidCFD-d...del/Make/files


the BC you want is commented out, so it's not compiled.

I tried uncommenting the BC you want, and compilation failed.

Sorry, but it is probably some work to get this to work. I have no idea if it is a little or a lot of effort.

Best regards,

Eric

Dear Eric/Foamers,


I managed to resolve this issue by adding the following as lib to controlDict:

 libs
    (
       "libturbulenceModels.so"
   )

Therefore, there is no error and the first iteration goes through just fine. But when it goes to the second iteration, things go wrong:

Time = 1


Solving for fluid region water
AINVPBiCG:  Solving for Ux, Initial residual = 1, Final residual = 0.000548682, No Iterations 3
AINVPBiCG:  Solving for Uy, Initial residual = 1, Final residual = 0.000573363, No Iterations 3
AINVPBiCG:  Solving for Uz, Initial residual = 1, Final residual = 0.000636619, No Iterations 3
AINVPBiCG:  Solving for h, Initial residual = 1, Final residual = 0.000759521, No Iterations 29
Min/max T:278.9 285.15
AINVPCG:  Solving for p_rgh, Initial residual = 0.99995, Final residual = 9.86198e-07, No Iterations 461
time step continuity errors : sum local = 0.166464, global = -0.00138159, cumulative = -0.00138159
Min/max rho:990 995.83
AINVPBiCG:  Solving for epsilon, Initial residual = 0.946346, Final residual = 0.000120173, No Iterations 4
bounding epsilon, min: 0 max: 3522.65 average: 0.0731319
AINVPBiCG:  Solving for k, Initial residual = 1, Final residual = 0.000111449, No Iterations 4
ExecutionTime = 23.36 s  ClockTime = 24 s

Time = 2


Solving for fluid region water
AINVPBiCG:  Solving for Ux, Initial residual = 1, Final residual = 0.000174815, No Iterations 4
AINVPBiCG:  Solving for Uy, Initial residual = 1, Final residual = 0.000173309, No Iterations 4
AINVPBiCG:  Solving for Uz, Initial residual = 0.999999, Final residual = 0.000178103, No Iterations 4
AINVPBiCG:  Solving for h, Initial residual = 1, Final residual = 0.000772795, No Iterations 30
Min/max T:-1.42592e+20 3.37913e+06
AINVPCG:  Solving for p_rgh, Initial residual = 1, Final residual = 9.64568e-07, No Iterations 460
time step continuity errors : sum local = 2.61101e+09, global = 1.98948e+06, cumulative = 1.98948e+06
Min/max rho:990 995.83
AINVPBiCG:  Solving for epsilon, Initial residual = 1, Final residual = 0.000766922, No Iterations 4
 bounding epsilon, min: 0 max: 4.24267e+14 average: 9.19178e+10
AINVPBiCG:  Solving for k, Initial residual = 1, Final residual = 0.000351024, No Iterations 4
ExecutionTime = 44.52 s  ClockTime = 45 s

Weird is that if I stop the computation after the first iteration and run it again (from the first iteration), then the second iteration goes through just fine:

Time = 1


Solving for fluid region water
AINVPBiCG:  Solving for Ux, Initial residual = 1, Final residual = 0.000548682, No Iterations 3
AINVPBiCG:  Solving for Uy, Initial residual = 1, Final residual = 0.000573363, No Iterations 3
AINVPBiCG:  Solving for Uz, Initial residual = 1, Final residual = 0.000636619, No Iterations 3
AINVPBiCG:  Solving for h, Initial residual = 1, Final residual = 0.000759521, No Iterations 29
Min/max T:278.9 285.15
AINVPCG:  Solving for p_rgh, Initial residual = 0.99995, Final residual = 9.86198e-07, No Iterations 461
time step continuity errors : sum local = 0.166464, global = -0.00138159, cumulative = -0.00138159
Min/max rho:990 995.83
AINVPBiCG:  Solving for epsilon, Initial residual = 0.946346, Final residual = 0.000120173, No Iterations 4
bounding epsilon, min: 0 max: 3522.65 average: 0.0731319
AINVPBiCG:  Solving for k, Initial residual = 1, Final residual = 0.000111449, No Iterations 4
ExecutionTime = 23.37 s  ClockTime = 23 s





Time = 2


Solving for fluid region water
AINVPBiCG:  Solving for Ux, Initial residual = 0.272723, Final residual = 0.000161735, No Iterations 3
AINVPBiCG:  Solving for Uy, Initial residual = 0.263371, Final residual = 0.000155051, No Iterations 3
AINVPBiCG:  Solving for Uz, Initial residual = 0.199896, Final residual = 7.86134e-05, No Iterations 3
AINVPBiCG:  Solving for h, Initial residual = 0.392765, Final residual = 0.000357602, No Iterations 24
Min/max T:278.9 285.151
AINVPCG:  Solving for p_rgh, Initial residual = 0.0593374, Final residual = 9.71312e-07, No Iterations 478
 time step continuity errors : sum local = 11.8758, global = 0.0295116, cumulative = 0.0295116
Min/max rho:990 995.83
AINVPBiCG:  Solving for epsilon, Initial residual = 0.82644, Final residual = 0.000255127, No Iterations 3
 bounding epsilon, min: 0 max: 29851.8 average: 0.273613
AINVPBiCG:  Solving for k, Initial residual = 0.720961, Final residual = 0.000259339, No Iterations 3
 ExecutionTime = 23.8 s  ClockTime = 24 s

The same procedure (if repeated) lets the simulation run fine. But I cannot stop and resume simulation after each iteration. Any ideas what might be causing this??


Any help is very much appreciated



PS: I acknowledge that there is a problem with epsilon; however, the issue with convergence if stopped/resumed puzzles me...

[edaymo]

Hi, MJ,

Thanks for the update. That was a clever solution. I am honestly not sure why the mapped T BC it's working when you include the turbulence model library ("libturbulenceModels.so"), since this BC is not supposed to be compiled in RapidCFD. But nonetheless, well done for getting it working.

However, I don't have a specific idea for your error. Could be a lot of causes, from the grid to a bad initial condition for your turbulence parameters to something RapidCFD (vs. OpenFOAM) related.

Does the same exact case run fine in OpenFOAM v2.3.1, or does the same problem occur in OpenFOAM? If yes, then it's more a general OpenFOAM issue. If no, then it's something related to RapidCFD in particular. This feedback might help with the determination of the root cause of your problems.

Best regards,

Eric