[TOC]

The new cluster, 'Valhalla', is replacing the current 'Potato' cluster. While the layout of the two servers is largely the same from a user's perspective, there are some important changes to note.

What makes this new infrastructure different?
---------------------------------------------

- One dedicated storage node
- 1 dedicated head node
- one head node for the graphical interface
- 1 TB memory and 96 CPUs compute machine
- 384 GB memory and 96 CPUs compute machine

Valhalla is a local dedicated cluster that allows the AFSG-MIB department to run computational analyses and store and share permanent and temporary files. It is suitable for small projects, as the number of available machines is limited. Larger projects should be run on the Dutch national supercomputer, Snellius or on the Wageningen tier 2 super computer Annuna.

The term 'cluster' refers to a computational infrastructure in which only the login nodes are interactive. Users can use the login nodes to perform basic operations and the compute nodes to run analyses.

Node overview
-------------

| Node type | CPU | #CPU/Threads | Clock speed (GHz) | Memory | Partition |
| :--- | :--- | :--- | :--- | :--- | :--- |
| Master | Intel Xeon D-2146NT | 8/16 | 3.0 | 128 GB | Small |
| Interactive | Intel Xeon D-2146NT | 8/16 | 3.0 | 128 GB | Small |
| Compute 01 | AMD EPYC 9654P | 96/192 | 3.71 | 995 GB | Bigmem |
| Compute 02 | AMD EPYC 9654P | 96/192 | 3.71 | 365 GB | Normal |

Operating system
----------------

The machines run Rocky Linux 10.

---------------------

[article_list depth:2]