Erlangen National Center for High Performance Computing (NHR@FAU)

Gauß-Allianz
Member
Go to first section

About Erlangen National Center for High Performance Computing (NHR@FAU)

On January 1, 2021, the Center for National High Performance Computing Erlangen (NHR@FAU) was established and, together with eight other centers, forms a national network. Erlangen thus forms a hotspot of scientific computing in northern Bavaria and improves the framework conditions for strengthening the entire Free State as a top location for high-performance computing. In addition to the investment in new hardware, the use of which will open up interesting new opportunities for scientists at FAU and throughout Germany, an interdisciplinary team will be established at the same time that will efficiently support the operation and use of the computer hardware. The northern Bavarian branch office of the “network of competence for scientific high performance computing in Bavaria” (KONWIHR) is in the meantime also located at NHR@FAU. FAU contributes to the national network in particular its established and internationally recognized competencies in complex simulations of atoms and molecules, hardware-efficient parallel programming, and its expertise in education and training. The cross-faculty activities in these fields are already bundled at FAU within the framework of the "Atomistic Structure Simulation Lab " and the "HPC Performance Lab ". RRZE is the central IT service provider of the FAU, ensuring an operative IT infrastructure for science, teaching, and management. It also supports other universities in Northern Bavaria within a regional service framework. Providing an interface between scientific research and technology, RRZE is a center of IT competence in the higher education landscape and a partner for science.

HPC Expertise

Model-based Performance Engineering
Courses with this focus are offered all over Germany and worldwide. For instance, RRZE HPC experts act as external trainers of the PATCs (PRACE Advanced Training Centers).

Sparse Matrix-Vector-Algorithms and the GHOST Library
Research activities at the RRZE have been ongoing since the year 2000 and are currently being conducted in the DFG priority programme “SPPEXA” under the “ESSEX” project.

Development of the LIKWID Multicore Toolsuite
LIKWID is an easy to install collection of tools that simplify the handling of multicore systems under the Linux OS. LIKWID can also be used as a monitoring backend and is already in operation in many computing centers of the GA. The development has started in 2009 and is currently being funded through the BMBF project “FEPA”.

HPC Systems

TinyGPU - Sysgen GPU-Nodes II, GPU-Nodes III, GPU-Nodes V, Supermicro 2124GQ-NART Nvidia GPU-Nodes IV TRADEX SYSTEMS Sp. z o.o. Gigabyte G481-H80

since October 2016 - Last Update: March 2021 - learn more
GPU nodes GPU system Tier 3 hpc system
45 Nodes
5.00 TiB
Main Memory
1,392 CPU Cores
Intel, AMD
208 GPGPUs
Nvidia
3 Applications
8 Versions

TinyFAT - Megware Supermicro X10DRH RaidMedia Gigabyte H262-Z66

since August 2016 - Last Update: November 2020 - learn more
47 Nodes
22.00 TiB
Main Memory
2,484 CPU Cores
Intel, AMD
HPC-System Alex

Alex - Megware GPGPU Cluster, GPU-Cluster, NF5488A5

since March 2022 - Last Update: March 2022 - learn more - Top500.org
GPU nodes GPU system Tier 2 hpc system Tier 3 hpc system
6.080 TFlop/s
Peak Performance
82 Nodes
Mellanox HDR200-Infiniband, HDR200-Infiniband
78.00 TiB
Main Memory
10,496 CPU Cores
AMD
656 GPGPUs
Nvidia

Meggie - Megware Linux-Cluster

since December 2016 - learn more - Top500.org
MPP system Tier 3 hpc system
511 TFlop/s
Peak Performance
728 Nodes
Intel Omnipath
46.00 TiB
Main Memory
14,560 CPU Cores
Intel
21 Applications
61 Versions
HPC-System Fritz

Fritz - Megware D50TNP

since March 2022 - Last Update: July 2023 - learn more - Top500.org
MPP system Tier 2 hpc system Tier 3 hpc system
5.450 TFlop/s
Peak Performance
992 Nodes
Mellanox HDR100-Infiniband
248.00 TiB
Main Memory
71,424 CPU Cores
Intel

File Systems

HOME file system

Global (Lenovo GPFS/NFS NFS)

since September 2020
HDD Symbol
50 TB
disk storage

WORK file system

Saturn (Hewlett Packard Enterprise (HPE) HPE Apollo 4510 Gen10 NFS)

since October 2018
HDD Symbol
450 TB
disk storage

Titan (Hewlett Packard Enterprise (HPE) HPE Apollo 4510 Gen10 NFS)

since April 2019
HDD Symbol
450 TB
disk storage

Fundus-Disk (Lenovo GPFS GPFS)

since June 2020
HDD Symbol
6000 TB
disk storage

Atuin (Dell PowerEdge XE7100/XE7440 NFS)

since January 2022
HDD Symbol
1000 TB
disk storage

Fritz (Megware Lustre Lustre)

since January 2022
HDD Symbol
3700 TB
disk storage

Archives

Fundus-Tape - since July 2020

  • 1x IBM TS4500
  • 8 tape drives
  • 3.370 magnetic cartridges
  • 15.000 TB tape storage

Contact

Zentrum für Nationales Hochleistungsrechnen Erlangen (NHR@FAU)
Martensstr. 1
91058 Erlangen

 Send a message to NHR@FAU

Legal Entity

Friedrich-Alexander-Universität Erlangen-Nürnberg

Mit rund 38.000 Studierenden, 242 Studiengängen, 4.000 Beschäftigten im wissenschaftlichen Bereich, darunter über 650 Professorinnen und Professoren, 160 Millionen Euro Drittmittel im Jahr und 500 Partnerschaften mit Universitäten in aller Welt gehört die 1743 gegründete Friedrich-Alexander-Universität Erlangen-Nürnberg (FAU) zu den großen, forschungsstarken und international ausgerichteten Universitäten in Deutschland. Im Mittelpunkt der forschungsorientierten Lehre steht die wissens- und methodenbasierte Ausbildung der Studierenden zu kritischen, eigenverantwortlichen Persönlichkeiten. Die Forschung selbst ist gleichermaßen grundlagen- wie anwendungsorientiert.

Contact

Friedrich-Alexander-Universität Erlangen-Nürnberg
Schlossplatz 4
91054 Erlangen
Germany