Table of Contents
Ancient Clusters
Please note, these are old clusters which are no longer in operation.
GPGPU cluster "Scout"
- Performance: 112 TFlop/s SP peek performance (9.3 TFlop/s DP)
- Hardware:
- 2 water-cooled racks,
- 108 GPU's, 432 GB GPU RAM,
- 27 NVIDIA Tesla S1070,
- 1 x Head Node: 8 CPU cores, 32 GB RAM,
- 18 x Compute Nodes: 8 CPU cores and 16 GB RAM per Node,
- 4 or 6 Tesla C1060 GPUs per node, 4 GB memory each,
- 7 TB disk capacity.
- Installed in 2008 on Campus Bockenheim.
Cluster III (1020 CPU cores)
- installed in 2006/07 on Campus Riedberg
- 4.1 TFLOPS performance
- 1020 CPU cores, 2.3 TB RAM, 80 TB disk capacity
- 9 air cooled racks
- Myrinet and Gigabit Ethernet interconnects
- 249 Nodes (185x Ethernet + 64x Myrinet)
- CPU: 64 bit Dual core AMD Opteron 270 2.0 GHz
- RAM: 44 nodes with 16 GB, all other nodes with 8 GByte RAM
- 2 frontend quad Opteron 846, 2.0 GHz, 8 GByte RAM
- 3 stroage nodes dual Opteron 248
- 5 TByte storage space each (S-ATA)
Cluster II (568 CPU's)
- Installed in 2004 on Campus Westend
- 2.0 TFLOPS performance
- 568 CPU's, 1.3 TB RAM, 27 TB disk capacity
- 10 water cooled racs
- Myrinet and Gigabit Ethernet interconnects
- 282 Nodes (218xEthernet+64xMyrinet)
- CPU: 64 bit Single core AMD Opteron DP 244 1,8GHz
- RAM: DIMM DDR 1GB ECC reg. PC333 Low Profile
- HDD: DiamondMax 740X 80GB 7200U 2MB 3,5“
- 1 MEGWARE Cluster Frontend Dual Opteron
- 1 MEGWARE Cluster CompileNode Dual Opteron