Hardware
Computers, laptops, displays and so on
Lothar is responsible for organising and maintaining all the hardware that you need for your day to day work. If you need anything specific (even if it's just a certain cable) feel free to ask him. He is also responsible for everything network related and can help you to access Asterope or Merope for the first time.
Asterope and Merope
The group owns two "clusters" (single nodes, actually) called Asterope and Merope. See the SSH and SCP page for access details. Both have two AMD EPYC 7763 processors with 64 cores each, 128GB of RAM and 17TB of local storage. Both run the Slurm Workload Manager to schedule the jobs.
Here you can find a cheat sheet with some useful Slurm commands: https://ircc.fiu.edu/download/user-guides/Slurm_Cheat_Sheet.pdf
Additional commands:
sfreeshow the numer of free cores.
magnitUDE and amplitUDE
magnitUDE
magnitUDE is a high-performance cluster operated by the university. Information on how to use it can be found here: https://www.uni-due.de/css/how_to_s
Performance data:
- Supplier: NEC Deutschland GmbH
- 624 Compute Nodes /14,976 Cores (since Jan. 2017)
- Intel E5-2650_v4 Broadwell CPUs
- Intel OmniPath Interconnect
- 56,064 GB Main Memory (since Jan. 2017)
- 480 TB NEC LXFS Storage
- High-Performance Linpack Benchmark (using 624 C-Nodes + 1 Login Node) - Rmax 489.3 TFlop/s - Rpeak 528.0 TFlop/s
amplitUDE
amplitUDE, the successor of magnitude, was supposed to start its test phase in the second half of 2023, but didn't, yet. For up to date information please check out https://www.uni-due.de/css/amplitude_main.php
Performance data (no guarantee):
- 240 CPU-nodes with 2x Intel® Xeon® Platinum 8480+ processor each, of which
- 176 "Standard" (512 GB RAM)
- 51 "High-Mem" (1 TB RAM)
- 13 "Super High-Mem"(2 TB RAM)
- 19 GPU-nodes with 2x Intel® Xeon® Platinum 8480+ processor each, of which
- 15 with 4x NVIDIA H100
- 4 with 2x NVIDIA H100
- Mellanox InfiniBand NDR200 with Blocking 1:1,66
- Lustre File-System with approx. 2,1 PB useable capacity
Training in High Performance Computing
- High-Performance Computing Center Stuttgart: https://www.hlrs.de/training/hpc-training
- Jülich Supercomputing Centre: https://www.fz-juelich.de/en/ias/jsc/education/training-courses