Infrastructure¶
The main components of the maxwell cluster are
- fast dedicated storage (cluster file-systems)
- fast, low-latency network (infiniband)
- high-memory compute nodes with substantial number of GPGPUs
The pages give a very brief overview on the corresponding components and their status (where possible).
- CPU+GPU nodes: 990
- Total number of cores with hyperthreading: 100656
- Total number of PHYSICAL cores: 50328
- Theoretical CPU peak performance: 1485 TFlops
- Total RAM: 600 TB
- GPU nodes: 195
- Total number of GPUs: 447
- Theoretical GPU peak performance: 4118 TFlops
- Total peak performance: 5600 TFlops
- GPFS exfel: 60 PB
- GPFS petra3: 30 PB
- GPFS cfel: 1.6 PB
- GPFS cssb: 11 PB
- DUST: 5 PB
- root switches: 6
- top switches: 12
- leaf switches: 42
- IB cables (#): ~1500
- IB cables (length): ~10km
- Users: ~4700
- Display Users: ~450 per day
- JupyterHub Users: ~570 per month
- Batch Jobs: ~550.000 per month
- Citations: >100 in 2025
Adding resources to maxwell¶
It's well possible to add for example groups resources to maxwell. Please get in touch maxwell.service@desy.de for details. Keep in mind that we will need to impose certain constraints to keep the cluster as little heterogeneous as feasible.