removed some links to data, added slurm script examples and readme

tutorials/hpc/05_your_turn/now_its_your_turn.py

@@ -12,13 +12,9 @@
 
 
 # "Asteroids": Asteroids of the Solar System
-# Use
-# ```
-# wget --no-check-certificate 'https://docs.google.com/uc?export=download&id=1E6mrMi7lL-UoHmZMTicVi0I2IdOKNn2-' -O ~/sbdb_asteroids.h5
-# ```
-# to download an example data set (92MB) consisting of the asteroids from the JPL Small Body Database (https://ssd.jpl.nasa.gov/tools/sbdb_lookup.html#/)
+# Download the data set of the asteroids from the JPL Small Body Database from https://ssd.jpl.nasa.gov/tools/sbdb_lookup.html#/
+# and load the resulting csv file to Heat.
 
-X = ht.load_hdf5("~/sbdb_asteroids.h5", device="gpu", dataset="data", split=0)
 
 # ... to be completed ...
 

tutorials/hpc/README.md

@@ -0,0 +1,12 @@
+There are two example scripts in this directory, `slurm_script_cpu.sh` and `slurm_script_gpu.sh`, that demonstrate how to run a Heat application on an HPC-system with SLURM as resource manager.
+
+1. `slurm_script_cpu.sh` is an example script that runs a Heat application on a CPU node. You must specify the name of the respective partition of your cluster. Moreover, the
+    numer of CPU cores available at a node of your system must be greater or equal to the product of the tasks-per-node- and the cpus-per-task-argument (=8x16=128 in the example).
+
+2. `slurm_script_gpu.sh` is an example script that runs a Heat application on a GPU node. You must specify the name of the respective partition of your cluster. Moreover, the
+    numer of GPU devices available at a node of your system must be greater or equal to the number of GPUs requested in the script (=4 in the example).
+
+## Remarks
+
+* Please have a look into the documentation of your HPC-system for its detailed configuration and properties. Maybe, you have to adjust the script to your system.
+* You need to load the required modules (e.g., for MPI, CUDA etc.) from the module system of your HPC-system before running the script. Moreover, you need to install Heat in a virtual environment (and activate it). Alternatively, you may use spack (if available on your system) for installing Heat and its dependencies.

tutorials/hpc/slurm_script_cpu.sh

@@ -0,0 +1,12 @@
+#!/bin/bash
+
+#SBATCH --partition=<partition_name>
+#SBATCH --nodes=1
+#SBATCH --tasks-per-node=8
+#SBATCH --cpus-per-task=16
+#SBATCH --time="00:01:00"
+
+export MKL_NUM_THREADS=$SLURM_CPUS_PER_TASK
+export OMP_NUM_THREADS=$SLURM_CPUS_PER_TASK
+
+srun python ~/heat/tutorials/hpc/01_basics/01_basics_dndarrays.py

tutorials/hpc/run_script.sh → tutorials/hpc/slurm_script_gpu.sh

@@ -1,10 +1,9 @@
 #!/bin/bash
 
-#SBATCH --partition=normal
-#SBATCH --reservation=haicon
+#SBATCH --partition=<partition_name>
 #SBATCH --nodes=1
 #SBATCH --tasks-per-node=4
-#SBATCH --cpus-per-task=12
+#SBATCH --cpus-per-task=2
 #SBATCH --gres=gpu:4
 #SBATCH --time="00:01:00"