Closed zingale closed 5 years ago
@dwillcox can you suggest a problem that uses the CUDA VODE for the test suite? Please give the make line.
Here are the 3 networks I've set up with CUDA VODE.
ignition_simple
Make line:
make -j COMP=PGI NDEBUG=t MPI= OMP= CUDA=t NETWORK_DIR=ignition_simple INTEGRATOR_DIR=VODE90 EOS_DIR=helmholtz BL_USE_F_BASELIB=t
Execute Line:
./main.Linux.PGI.exe inputs_ignition.32
ignition_reaclib/URCA-simple
Make Line:
make -j COMP=PGI NDEBUG=t MPI= OMP= CUDA=t NETWORK_DIR=ignition_reaclib/URCA-simple INTEGRATOR_DIR=VODE90 EOS_DIR=helmholtz BL_USE_F_BASELIB=t
Execute Line:
./main.Linux.PGI.exe inputs_reaclib_urca.32
aprox13
Make Line:
make -j COMP=PGI NDEBUG=t MPI= OMP= CUDA=t NETWORK_DIR=aprox13 INTEGRATOR_DIR=VODE90 EOS_DIR=helmholtz BL_USE_F_BASELIB=t
Execute Line:
./main.Linux.PGI.exe inputs_aprox13.32
amrex branch is gpu
Microphysics branch is cudadevice
ah... I think I need to wait until it is all in development
there is now a nightly GPU test suite
We have no GPU tests in the PGI test suite. We should pick some basic tests to add coverage.