Dear all-Users:
I compiled the lammps (Aug2024) by using the following commands:
cmake …/cmake/ -C …/cmake/presets/basic.cmake -C …/cmake/presets/gpu-cuda.cmake -D GPU_API=cuda -D GPU_ARCH=sm_89
make -j 30;
For the basic.cmake, I just modified the following line:
set(ALL_PACKAGES GPU KSPACE MANYBODY MOLECULE MEAM REAXFF)
After getting the lmp, I run my simulation by: “mpirun -np 1 ./lmp -sf gpu < in.test”. The inpute file can be see below.
in.test (1.6 KB)
There is no error and the GPU Util is very small, as can be seen below.
BTW, the GPU is working for other potentials such as tersoff and lj (simulation is very fast).
Is there someone knows the reason? Thanks so much.