FIX: runtime docu for Gromacs
All checks were successful
Build and Deploy Documentation / build-and-deploy (push) Successful in 7s

This commit is contained in:
2025-10-08 13:17:13 +02:00
parent a1b3941c62
commit 55fcd1b665

View File

@@ -33,22 +33,51 @@ module load gcc/12.3 openmpi/5.0.7-3vzj-A100-gpu gromacs/2025.2-vbj4-A100-gpu-om
module use Spack unstable module use Spack unstable
module load gcc/12.3 openmpi/5.0.7-blxc-GH200-gpu gromacs/2025.2-cjnq-GH200-gpu-omp module load gcc/12.3 openmpi/5.0.7-blxc-GH200-gpu gromacs/2025.2-cjnq-GH200-gpu-omp
``` ```
### SBATCH A100, 4 GPU, 16 OMP threads, 4 MPI ranks
### SBATCH GH, 4 GPU, 32 OMP threads, 4 MPI ranks
```bash ```bash
#!/bin/bash #!/bin/bash
#SBATCH --get-user-env #SBATCH --time=00:10:00 # maximum execution time of 10 minutes
#SBATCH --output=_scheduler-stdout.txt #SBATCH --output=_scheduler-stdout.txt
#SBATCH --error=_scheduler-stderr.txt #SBATCH --error=_scheduler-stderr.txt
#SBATCH --job-name="Testing GROMACS GH"
#SBATCH --nodes=1 # number of GH200 nodes with each node having 4 CPU+GPU #SBATCH --nodes=1 # number of GH200 nodes with each node having 4 CPU+GPU
#SBATCH --ntasks-per-node=4 # 8 MPI ranks per node #SBATCH --ntasks-per-node=4 # 4 MPI ranks per node
#SBATCH --cpus-per-task 32 # 32 OMP threads per MPI rank #SBATCH --cpus-per-task=16 # 16 OMP threads per MPI rank
#SBATCH --cluster=gmerlin7 #SBATCH --cluster=gmerlin7
#SBATCH --hint=nomultithread #SBATCH --hint=nomultithread
#SBATCH --partition=gh-hourly #SBATCH --partition=a100-hourly
#SBATCH --gpus=4 #SBATCH --gpus=4
#SBATCH --gpus-per-task=1
unset PMODULES_ENV
module purge
module use Spack unstable
module load gcc/12.3 openmpi/5.0.7-3vzj-A100-gpu gromacs/2025.2-vbj4-A100-gpu-omp
export FI_CXI_RX_MATCH_MODE=software
export GMX_GPU_DD_COMMS=true
export GMX_GPU_PME_PP_COMMS=true
export GMX_FORCE_UPDATE_DEFAULT_GPU=true
export GMX_ENABLE_DIRECT_GPU_COMM=1
export GMX_FORCE_GPU_AWARE_MPI=1
# Add your input (tpr) file in the command below
srun gmx_mpi grompp -f step6.0_minimization.mdp -o step6.0_minimization.tpr -c step5_input.gro -r step5_input.gro -p topol.top -n index.ndx
srun gmx_mpi mdrun -s step6.0_minimization.tpr -ntomp ${SLURM_CPUS_PER_TASK}
```
### SBATCH GH, 2 GPU, 18 OMP threads, 2 MPI ranks
```bash
#!/bin/bash
#SBATCH --time=00:10:00 # maximum execution time of 10 minutes
#SBATCH --output=_scheduler-stdout.txt
#SBATCH --error=_scheduler-stderr.txt
#SBATCH --nodes=1 # number of GH200 nodes with each node having 4 CPU+GPU
#SBATCH --ntasks-per-node=2 # 2 MPI ranks per node
#SBATCH --cpus-per-task=18 # 18 OMP threads per MPI rank
#SBATCH --cluster=gmerlin7
#SBATCH --hint=nomultithread
#SBATCH --partition=gh-hourly
#SBATCH --gpus=2
unset PMODULES_ENV unset PMODULES_ENV
module purge module purge
@@ -63,7 +92,9 @@ export GMX_FORCE_UPDATE_DEFAULT_GPU=true
export GMX_ENABLE_DIRECT_GPU_COMM=1 export GMX_ENABLE_DIRECT_GPU_COMM=1
export GMX_FORCE_GPU_AWARE_MPI=1 export GMX_FORCE_GPU_AWARE_MPI=1
srun gmx_mpi mdrun -s input.tpr -ntomp 32 -bonded gpu -nb gpu -pme gpu -pin on -v -noconfout -dlb yes -nstlist 300 -npme 1 -nsteps 10000 -update gpu # Add your input (tpr) file in the command below
srun gmx_mpi grompp -f step6.0_minimization.mdp -o step6.0_minimization.tpr -c step5_input.gro -r step5_input.gro -p topol.top -n index.ndx
srun gmx_mpi mdrun -s step6.0_minimization.tpr -ntomp ${SLURM_CPUS_PER_TASK}
``` ```
## Developing your own GPU code ## Developing your own GPU code
@@ -74,7 +105,7 @@ module use Spack unstable
module load gcc/12.3 openmpi/5.0.7-3vzj-A100-gpu gromacs/2025.2-vbj4-A100-gpu-omp cmake/3.31.6-o3lb python/3.13.1-cyro module load gcc/12.3 openmpi/5.0.7-3vzj-A100-gpu gromacs/2025.2-vbj4-A100-gpu-omp cmake/3.31.6-o3lb python/3.13.1-cyro
git clone https://github.com/gromacs/gromacs.git git clone https://github.com/gromacs/gromacs.git
cd gromacs cd gromacs
mkdir build && cd build mkdir build && cd build
cmake -DCMAKE_C_COMPILER=gcc-12 \ cmake -DCMAKE_C_COMPILER=gcc-12 \