Compute nodes are grouped into partitions to allow the user to select different hardware to run their software on. Each partition includes a subset of nodes with a different type of hardware and a specific maximum wall time and minimal number of CPUs.


This page sumarizes the partitions available to users and the corresponding accounting weights. 

This page assumes knowledge on partition usage and how to submit a job using SLURM. Please refer to the accounting documentation and  HPC user guide for a general introduction to these topics.


Snellius partitions

Compute nodes are grouped into partitions to allow the user to select different hardware to run their software on. Each partition includes a subset of nodes with a different type of hardware and a specific maximum wall time.

The partitions can be selected by users via the SLURM option:

#SBATCH --partition=<partition name>

or its short form:

#SBATCH -p <partition name>

The partitions available on Snellius are summarised in the table below. For details of the different hardware available on each node, please look at the Snellius hardware page.




Partition nameType SpecNode descriptionNumber of nodes
CPU Cores
per node		GPUs
per node		Available GiB
RAM /  node 		Smallest possible
allocation

SBU weight

SBU / CPU hour
or
SBU / GPU hour

rome CPU thin compute node
AMD Rome CPU		521128-2241/8 node:
16 cores +
28 GiB RAM		1.0
genoa CPU thin compute node
AMD Genoa CPU		737192-3361/8 node:
24 cores +
42 GiB RAM		1.0
fat_rome CPU FATfat compute node
AMD Rome CPU
72128-9601/8 node:
16 cores + 
120 GiB RAM		1.5
fat_genoa CPU FATfat compute node
AMD Genoa CPU
48192-14401/8 node:
24 cores +
180 GiB RAM		1.5
himem_4tb CPU HIMEMhigh memory node 

2128-38401/8 node:
16 cores +
480 GiB RAM		2.0
himem_8tb CPU HIMEMhigh memory node2128-76801/8 node:
16 cores +
960 GiB RAM		3.0
gpu_a100 GPU NVIDIA A100

63

7244801/4 node:
18 cores + 1 GPU +
120 GiB RAM		128
gpu_h100 GPU NVIDIA H10088644720 1/4 node:
16 cores + 1 GPU
+ 180 GiB RAM (host)		192
gpu_mig GPU MIGNVIDIA A100 MIG
(Multi-Instance GPU)		4728*480 1/8 node:
9 cores + 1 GPU (MIG) +
60 GiB RAM		64
gpu_vis GPU visualizationNVIDIA A100
for visualization		637244801/4 node:
18 cores + 1 GPU +
120 GiB RAM		128
staging service SMTSMT 32 threads1016*-2241 thread + 
7 GiB RAM		2.0
cbuild service SMTSMT 32 threads1016*-2241 thread + 
7 GiB RAM		2.0


Memory

The “Available memory per node” is the amount of memory available to users and what can be requested within a job. This value is smaller than the “Total memory” on the node, as it doesn't include the memory reserved for the OS and other system's processes.  

Walltime

The walltime, the maximum amount of real elapsed time a job is allowed to run before it is automatically stopped by the scheduler, is 120 hours for all partitions. One exception is the gpu_vis partition, where the maximum walltime is 24 hours.

Number of nodes 

Some nodes may be temporarily removed from or added to the system for maintenance and upgrades; therefore, the number of available nodes is not constant.

Node names

Node names for each partitions can be determined with the sinfo command

Short jobs (1 hour walltime)

Whenever you submit a job that uses at most 1 hour walltime to the “thin”, “fat” or “gpu” partitions, SLURM will schedule the job on a node that is only available for such short jobs. This effectively reduces the wait time for short jobs compared to longer jobs, which is useful for testing the setup and correctness of your jobs before submitting long-running production runs.

Note that the number of nodes that can run short jobs is relatively small. So submitting a short running job which uses many (e.g. tens or hundreds of) nodes will not work