Lichtenberg II Cluster Darmstadt

Lichtenberg II Cluster Darmstadt

Cluster Access

University

Technische Universität Darmstadt

Info

Like its predecessor, the successor to the Lichtenberg cluster is a high-performance computer in the medium performance class (Tier 2). It has a heterogeneous architecture including large memory nodes, accelerator systems with dedicated NVIDIA GP-GPUs and several DGX A100 systems to support modern AI research.

Add. Info

HPC consultation hour an cluster introduction:

  • Video Consultation for all questions about using the cluster and user and project applications: every Wednesday 10 – 11 am as a Zoom meeting:
    Meeting-ID: 665 7163 2291
    Kenncode: 113329
  • Monthly introduction to the use of the cluster: usually every 2nd Tuesday of the month
HPC Support
HKHLR Team Technische Universität Darmstadt
Downloads
Quick Reference Card Lichtenberg Darmstadt
Cluster Access

The cluster is open to scientists from all universities and public research institutions in Germany. The computing resources are granted via a science-led allocation procedure.

Typical Node Parameters

Title
MPI section
Cores (sockets x cores/socket)
2x48
Memory
384 GB
FLOPS/Core (DP, theor. peak)
52 GFLOPS
CPU Type
Intel Xeon Platinum 9242
Bandwidth
100 GB/s
Memory Bandwidth (triad)
400 GB/s
Title
MPI section 2
Cores (sockets x cores/socket)
2x52
Memory
512 GB
CPU Type
Intel Xeon Platinum 8470Q
Bandwidth
100 GB/s
Memory Bandwidth (triad)
1000 GB/s
Title
MEM section
Cores (sockets x cores/socket)
2x48
Memory
1536 GB
FLOPS/Core (DP, theor. peak)
52 GFLOPS
CPU Type
Intel Xeon Platinum 9242
Bandwidth
100 GB/s
Memory Bandwidth (triad)
400 GB/s
Title
MEM section 2
Cores (sockets x cores/socket)
2x52
Memory
2048 GB
CPU Type
Intel Xeon Platinum 8470Q
Bandwidth
100 GB/s
Memory Bandwidth (triad)
1000 GB/s
Title
ACC section GPU
Cores (sockets x cores/socket)
2x48
Memory
384 GB
CPU Type
Intel Xeon Platinum 8260
Bandwidth
100 GB/s
Memory Bandwidth (triad)
400 GB/s
Accelerators

4x 4x Nvidia Volta V100

4x 4x Nvidia Ampere A100

3x 8x Nvidia Ampere A100

Title
ACC section GPU 2
Cores (sockets x cores/socket)
2x52
Memory
512 GB
CPU Type
Intel Xeon Platinum 8470Q
Bandwidth
100 GB/s
Memory Bandwidth (triad)
1000 GB/s
Accelerators

2x 4x Nvidia Hopper H100

4x 4x Intel Ponte Vecchio GPU Max 1550

2x 8x AMD MI300X

Node Allocation
Shared and exclusive

Global Cluster Parameters

Processors (CPU, DP, peak)
ca. 8.5 PFLOPS
Accelerators (GPU, DP, peak)
1.7 PFLOPS
Computing cores (CPU)
122.000
Job Manager
Slurm Workload Manager
Other Job Constraints

runtime: 24h, max. 7d

Related Projects

Participating Universities