This shows you the differences between two versions of the page.
Next revision | Previous revision | ||
howtouseslurm [2019/05/11 18:58] – created tmatejuk | howtouseslurm [2023/08/01 01:08] (current) – external edit 127.0.0.1 | ||
---|---|---|---|
Line 1: | Line 1: | ||
- | You can use slurm in two ' | ||
- | ===Batch mode=== | + | [[howtouseslurm_000|SLURM, |
- | Most computation is done with batch mode. We prepare batch script (job). This script contains declaration of resources we require, and launch of computation we would like to run on them. Then we commit this batch script with '' | + | |
+ | [[howtouseslurm_001|SLURM, | ||
- | ---- | + | [[howtouseslurm_002|SLURM, |
- | ===Interactive mode=== | + | [[howtouseslurm_004|SLURM, how to build pipelines]] |
- | Its like login ( with ssh ) to computing node, but we can decide ( with SLURM ) how much resources we require. And unlike with ssh, SLURM makes sure we get resources nobody else is already using. | + | |
- | For example if we need interactive session with resources of 4 cores, and 200GB memory we use command : | + | [[howtouseslurm_003|SLURM, good to know]] |
- | srun -N1 -n4 --mem=200G --pty bash -l | + | |
- | Or if we need interactive session with resources of 1 core and 1 GPU card : | + | [[https:// |
- | srun -N1 -n1 --gres=gpu: | + | |
- | + | ||
- | Remember to use '' | + | |