Differenze

Queste sono le differenze tra la revisione selezionata e la versione attuale della pagina.

--- calcoloscientifico:userguide:gromacs [25/09/2023 13:05] – fabio.spataro
+++ calcoloscientifico:userguide:gromacs [20/08/2025 18:00] (versione attuale) – [Job Gromacs CPU] fabio.spataro
@@ Linea 22: / Linea 22: @@
 ^ Gromacs  ^ Plumed  ^ Compiler   ^ MPI             ^ CUDA SDK  ^ CUDA RTL  ^ CUDA Architectures  ^
 | 4.5.7    |         | GNU 5.4.0  | OpenMPI 1.10.7  |           |           |                     |
-| 5.1.4    | 2.4.8   | GNU 5.4.0  | OpenMPI 1.10.7  |           |           |                     |
+| 5.1.4    | 2.3.8   | GNU 5.4.0  | OpenMPI 1.10.7  |           |           |                     |
-| 5.1.5    | 2.4.8   | GNU 5.4.0  | OpenMPI 1.10.7  |           |           |                     |
+| 5.1.5    | 2.3.8   | GNU 5.4.0  | OpenMPI 1.10.7  |           |           |                     |
 | 2016.4   | 2.3.8   | GNU 5.4.0  | OpenMPI 1.10.7  | 10.0.130  | 10.0.130  | 6.0, 7.0            |
 | 2016.6   | 2.3.8   | GNU 5.4.0  | OpenMPI 1.10.7  | 10.0.130  | 10.0.130  | 6.0, 7.0            |
@@ Linea 512: / Linea 512: @@
 mpirun -np 4 gmx mdrun -deffnm meta -dlb auto -plumed plumed.dat
+</code>
+===== Gromacs with Apptainer =====
+Available versions of Gromacs:
+^ Gromacs                                                      ^
+| [[http://manual.gromacs.org/documentation/2025.1|2025.1]]    |
+==== Examples ====
+The following examples demonstrate using the [[https://catalog.ngc.nvidia.com/orgs/hpc/containers/gromacs|NGC GROMACS]] container to run the ''STMV'' benchmark.
+Download the ''STMV'' benchmark:
+<code bash>
+wget https://zenodo.org/record/3893789/files/GROMACS_heterogeneous_parallelization_benchmark_info_and_systems_JCP.tar.gz
+tar xf GROMACS_heterogeneous_parallelization_benchmark_info_and_systems_JCP.tar.gz
+cd GROMACS_heterogeneous_parallelization_benchmark_info_and_systems_JCP/stmv
+</code>
+==== Job Gromacs CPU ====
+Script ''slurm-gromacs-2025.1-cpu.sh'' to request one node and start 8 thread-MPI tasks and 6 OpenMP threads for each thread-MPI task (48 threads overall) and run the ''STMV'' benchmark:
+<code bash slurm-gromacs-2025.1-cpu.sh>
+#!/bin/bash --login
+#SBATCH --job-name=mdrun_cpu
+#SBATCH --output=%x.o%j
+#SBATCH --error=%x.e%j
+#SBATCH --nodes=1
+#SBATCH --ntasks-per-node=8
+#SBATCH --cpus-per-task=6
+#SBATCH --time=0-04:00:00
+#SBATCH --mem=16G
+#SBATCH --partition=cpu_guest
+#SBATCH --qos=cpu_guest
+##SBATCH --account=<account>
+shopt -q login_shell || exit 1
+test -n "$SLURM_NODELIST" || exit 1
+test $SLURM_NNODES -eq 1 || exit 1
+module load apptainer
+module load gromacs/2025.1/cpu
+apptainer run \
+    --bind "$PWD:/host_pwd" \
+    --pwd /host_pwd \
+    "$GROMACS_CONTAINER" \
+    gmx mdrun \
+        -v \
+        -ntmpi $SLURM_TASKS_PER_NODE \
+        -ntomp $SLURM_CPUS_PER_TASK \
+        -nb cpu \
+        -pme cpu \
+        -npme 1 \
+        -update cpu \
+        -bonded cpu \
+        -nsteps 100000 \
+        -resetstep 90000 \
+        -noconfout \
+        -dlb no \
+        -nstlist 300 \
+        -pin on
+</code>
+==== Job Gromacs GPU ====
+Script ''slurm-gromacs-2025.1-gpu.sh'' to request one node and start 8 thread-MPI tasks and 6 OpenMP threads for each thread-MPI task (48 threads thread) with 8 CUDA GPU's and run the ''STMV'' benchmark:
+<code bash slurm-gromacs-2025.1-gpu.sh>
+#!/bin/bash --login
+#SBATCH --job-name=mdrun_gpu
+#SBATCH --output=%x.o%j
+#SBATCH --error=%x.e%j
+#SBATCH --nodes=1
+#SBATCH --ntasks-per-node=8
+#SBATCH --cpus-per-task=6
+#SBATCH --time=0-00:30:00
+#SBATCH --mem=8G
+#SBATCH --gres=gpu:a5000_mm1:8
+#SBATCH --partition=gpu_guest
+#SBATCH --qos=gpu_guest
+##SBATCH --account=<account>
+shopt -q login_shell || exit 1
+test -n "$SLURM_NODELIST" || exit 1
+test $SLURM_NNODES -eq 1 || exit 1
+module load apptainer
+module load gromacs/2025.1/gpu
+apptainer run \
+    --nv \
+    --bind "$PWD:/host_pwd" \
+    --pwd /host_pwd \
+    "$GROMACS_CONTAINER" \
+    gmx mdrun \
+        -v \
+        -ntmpi $SLURM_TASKS_PER_NODE \
+        -ntomp $SLURM_CPUS_PER_TASK \
+        -nb gpu \
+        -pme gpu \
+        -npme 1 \
+        -update gpu \
+        -bonded gpu \
+        -nsteps 100000 \
+        -resetstep 90000 \
+        -noconfout \
+        -dlb no \
+        -nstlist 300 \
+        -pin on
 </code>