Initial commit

.gitignore

+*.log

README.md

+# Slurm
+
+Slurm is a workload manager, it runs on: `slurm.doc.gold.ac.uk`. Slurm
+can be accessed via ssh using your usual college credentials – the
+same as for igor.
+
+To run your code on a GPU compute node you first need to copy all of
+your code to your DOC shared home directory and then submit it as a
+job to Slurm.
+
+For example, to run this demo, ssh into `slurm.doc.gold.ac.uk` and
+clone this repository.
+
+``` sh
+ssh <user>@slurm.doc.gold.ac.uk
+git clone https://gitlab.doc.gold.ac.uk/slurm/mnist-slurm-demo.git
+cd mnist-slurm-demo
+sbatch mnist-job.sh             # This command submits the job.
+```
+
+Slurm will schedule your job immediately when resources become
+available.
+
+## Example batch script
+
+``` sh
+#!/bin/sh
+#SBATCH --time 4           # time in minutes to reserve
+#SBATCH --cpus-per-task 2  # number of cpu cores
+#SBATCH --mem 4G           # memory pool for all cores
+#SBATCH --gres gpu:1       # number of gpu cores
+#SBATCH  -o mnist.log      # write output to log file
+
+# Run jobs using the srun command.
+srun -l python mnist.py
+```
+
+# Useful commands
+
+See the [documentation](https://slurm.schedmd.com) for more
+information.
+
+## Submitting jobs
+
+``` sh
+sbatch mnist-job.sh
+```
+
+## Show information about a running job
+
+Replace `<n>` below with your job id.
+
+``` sh
+sstat <n>
+```
+
+## Show information about all submitted jobs
+
+``` sh
+sacct
+```
+
+## Show specific information about a job
+
+``` sh
+sacct --units G --format=jobid,avecpu,alloccpus,avevmsize -j <n>
+```

mnist-job.sh

+#!/bin/sh
+#SBATCH --time 4           # time in minutes to reserve
+#SBATCH --cpus-per-task 2  # number of cpu cores
+#SBATCH --mem 4G           # memory pool for all cores
+#SBATCH --gres gpu:1       # number of gpu cores
+#SBATCH  -o mnist.log      # write output to log file
+
+# Run jobs using the srun command.
+srun -l python mnist.py

mnist.py

+import tensorflow as tf
+mnist = tf.keras.datasets.mnist
+
+(x_train, y_train), (x_test, y_test) = mnist.load_data()
+x_train, x_test = x_train / 255.0, x_test / 255.0
+
+model = tf.keras.models.Sequential([
+  tf.keras.layers.Flatten(input_shape=(28, 28)),
+  tf.keras.layers.Dense(512, activation=tf.nn.relu),
+  tf.keras.layers.Dropout(0.2),
+  tf.keras.layers.Dense(10, activation=tf.nn.softmax)
+])
+model.compile(optimizer='adam',
+              loss='sparse_categorical_crossentropy',
+              metrics=['accuracy'])
+
+model.fit(x_train, y_train, epochs=5)
+print('Evaluation:', model.evaluate(x_test, y_test))