Closed timfelle closed 7 months ago
Seems the problem is with the projection. Did you successfully run the case without any projections? (Currently the projection is spurious for both CPU and GPU, and I am currently working on it)
I only tested whether the examples we supply will run without any modification, which they should, so i guess we should just turn off the projections for this case?
It's just my suspicion that the problem is from projections. I have no idea whether this case works or not without projections. Could you please test it once? If it works, then we should just turn off the projections and turn on it again once the projection is fixed.
Disabling the projections worked
The lid example does not complete with the case file supplied in the example.
-------Job Information--------
Start time: 06:52 / 2024-03-04 Running on: 1 MPI rank CPU type : Intel(R) Xeon(R) Gold 6226R CPU @ 2.90GHz Bcknd type: Accelerator (CUDA) Dev. name : NVIDIA A100-PCIE-40GB Real type : double precision
Last time step
``` ---------------------------------------------------------------- t = 0.4693600E+02 [ 93.87% ] ---------------------------------------------------------------- Time-step: 46937 CFL: 0.3994580+165 dt: 0.1000000E-02 ------------Fluid------------- Projection Pressure Proj. vec.: Orig. residual: 5 Infinity *** WARNING: New vector not linearly indepependent! *** *** WARNING: New vector not linearly indepependent! *** *** WARNING: New vector not linearly indepependent! *** *** WARNING: New vector not linearly indepependent! *** Pressure Iterations: Start residual: Final residual: 1 Infinity 0.0000000E+00 X-Velocity Iterations: Start residual: Final residual: 201 NaN NaN Y-Velocity Iterations: Start residual: Final residual: 201 NaN NaN Z-Velocity Iterations: Start residual: Final residual: 201 NaN NaN *** ERROR: Fluid solver diverged *** ```