skip breadcrumb navigation to main content

Bielefelder IT-Servicezentrum

Glasfaserkabel in einem Rechenzentrum — © Universität Bielefeld | BITS

Direct to

Bielefeld GPU cluster

The Faculty of Physics at Bielefeld University has procured a new high-performance computer based on GPUs in 2019. Due to a co-funding of the cluster by the university, a part of the GPU resources can also be made available to researchers of other faculties.

Scope of services

Provision of resources on the Bielefeld GPU Cluster for use by GPU (CUDA/openCL) optimized computing applications.

Technical details of the Bielefeld GPU Cluster:

Number of GPU Nodes: 28
Number of GPUs: 224 NVIDA Tesla V100 (32 GB)
Number of CPUs: 56 10-Core Intel Xeon
CPU memory: 7.2 TB total
GPU memory: 10.8 TB total
Storage: 2 PB parallel file system (BeeGFS)

Requirements

Target group

Employees

Request

Cluster usage can be requested by emailing gpucluster@uni-bielefeld.de.

Brief description of the project with justification for the use of GPU resources.
Project ending date
Specification of the computing time requirement in GPU hours
Specification of the storage requirements
Software requirements

Deployment times

1 working day from commissioning.

Duty to cooperate

Licensing of the application software by the faculty/department
Windows clients in university-wide AD
Standard hardware (according to framework agreement)

Support

Mon.-Fri. 09:00-17:00
Email: gpucluster@physik.uni-bielefeld.de