Free Slip Boundary Condition on GPUs

[gdhir]

Hi,

I wanted to know if the free slip boundary condition has been implemented on the GPU. As soon as I used the function “setSlipBoundary(sLattice, superGeometry, 2);”, I received a runtime error message displayed as follows:

terminate called after throwing an instance of ‘std::runtime_error’
what(): Legacy post processors not supported on GPU_CUDA

Is there a way to avoid this error while using the free slip boundary condition on the GPU. If not, is there a way to modify the current free slip code to avoid this error. I guess possible solutions might be to have it run on the GPU completely or to run the free slip code on the CPU somehow and rest of the code on GPU.

Any thoughts or suggestions will be very valuable.

[Adrian]

You are probably not using the latest version of OpenLB. In 1.7 the slip boundaries were ported into the GPU-supporting style. You can either switch to that release or backport the BC into 1.5/6. I recommend to switch to 1.7.

[gdhir]

Thanks for your reply. I have been able to find the free slip boundary code which is adapted to GPU operator style in v1.7. I have successfully added/ported the same to v1.6 and have been able to run a successful simulation on the GPU for my current usecase. Porting was needed since I am currently having several other custom integrations with v1.6.

Thanks again for your help with the same.

[Author]

Posts