25 May12:46pm
Hermes Cluster: The cluster is unit 27 of NANBIOSIS, HERMES has more than 3260 cores and 5920 parallel processing threads, 26TB of RAM, 264TB nvme and 400TB HDD of shared storage, all connected by a 100Gbps backbone network. In addition, we currently have 251,392 cuda cores, 720GB graphics memory, 1,179.42 peak FP16 TFLOPS, and 828.6 peak FP32 TFLOPS. HERMES is capable of processing a year’s worth of work on a personal computer in just 3 hours, greatly accelerating the work of researchers. Technical staff maintain the system and offer technical support, advice on parallel and high-performance supercomputing.
Highlighted figures:
- +3,260 Cores
- +5920 Threads
- 26TB RAM
- 264TB NVMe
- 400TB HDD
- 100 Gbps backbone
- 251,392 CUDA cores
- 720 GB Graphics memory