Introduction¶

About Merlin7¶

PSI's Merlin7 cluster is run on top of an IaaS (Infrastructure as a Service) vCluster on the CSCS Alps infrastructure. It is fully integrated with the PSI service landscape was designed to provide the same end user experience as its PSI-local predecessor clusters.

Merlin7 has been in production since beginning of June 2025.

All PSI users can request access to Merlin7, please go to the Requesting Merlin Accounts page and complete the steps given there.

In case you identify errors or missing information, please contact us directly or submit a ticket using the PSI service portal.

Infrastructure¶

Overview¶

A condensed overview of the Merlin cluster is shown in the diagram below:

architecture-beta
    service wan(internet)[Outside]
    group merlin(cloud)[Merlin7]
        service login_a(server)[login001] in merlin
        service login_b(server)[login 002] in merlin

        service cpu(server)[CPU CNs] in merlin
        service gpu(server)[GPU CNs] in merlin

    group slurm(server)[SLURM]
        service ctl(server)[Controller] in slurm
        service db(server)[Database] in slurm

    service storage(disk)[Storage]

    junction center in merlin
    junction ja in merlin
    junction jb in merlin

    wan:R <-- L:center
    center:T --> B:login_a
    center:B --> T:login_b

    login_a:R -- L:ja
    login_b:R -- L:jb
    ja:R --> L:cpu
    jb:R --> L:gpu
    ja:R -- L:gpu
    jb:R -- L:cpu

    cpu:B <--> T:gpu

    ctl{group}:B <--> T:login_a{group}
    cpu{group}:T <--> B:storage

    align row wan center
    align row ctl db storage
    align row login_a ja cpu
    align row login_b jb gpu
    align column ctl login_a center login_b
    align column storage cpu gpu

From the "outside" (either PSI intranet, VPN, or hopx gateway) a user can connect to the login nodes. From these a user can submit SLURM jobs to the compute nodes (both CPU and GPU nodes) or access them directly through an interactive session.

The SLURM controller manages the jobs and stores information about them in a database.

The Merlin cluster uses networked storage server for all nodes.

Hardware¶

The Merlin7 cluster contains the following node specification:

Node	#N	CPU	RAM	GPU	#GPUs
Login	2	2 AMD EPYC 7742 (64 Cores 2.25GHz)	512GB
CPU	77	2 AMD EPYC 7742 (64 Cores 2.25GHz)	512GB
GPU A100	5	2 AMD EPYC 7713 (64 Cores 3.2GHz)	512GB	A100 80GB	4
GPU GH	3	NVIDIA ARM Grace Neoverse v2 (144 Cores 3.1GHz)	864GB (Unified)	GH200 120GB	4

Network¶

The Merlin7 cluster builds on top of HPE/Cray technologies, including a high-performance network fabric called Slingshot. This network fabric is able to provide up to 200 Gbit/s throughput between nodes. Further information on Slignshot can be found on at HPE.

Through software interfaces like libFabric (which available on Merlin7), application can leverage the network seamlessly.

Storage¶

Unlike previous iteration of the Merlin HPC clusters, Merlin7 does not have any local storage. Instead storage for the entire cluster is provided through a dedicated storage appliance from HPE/Cray called ClusterStor.

The appliance is built of several storage servers:

2 management nodes
2 MDS servers, 12 drives per server, 2.9TiB (Raid10)
8 OSS-D servers, 106 drives per server, 14.5 T.B HDDs (Gridraid / Raid6)
4 OSS-F servers, 12 drives per server 7TiB SSDs (Raid10)

With effective storage capacity of:

10 PB HDD
- value visible on linux: HDD 9302.4 TiB
162 TB SSD
- value visible on linux: SSD 151.6 TiB
23.6 TiB on Metadata

The storage is directly connected to the cluster (and each individual node) through the Slingshot NIC.