CESGA offers HPC computing and storage resources and support to create different solutions.
SERVICE DESCRIPTION
CESGA will provide access to HPC resources. Finisterrae-II supercomputer with more than 306 nodes, each with 24 cores Haswell 2680v3 and 128GB of RAM. It is attached to a LUSTRE filesystem with more than 768TB of disk. The low latency network is Infiniband FDR@56Gbps. It includes additional nodes with special characteristics as GPU (8 NVIDIA K80), Xeon Phi (4) and one node wih 2TB RAM and 8 Intel Haswell 8867v3, 128 cores.
This infrastructure is a Tier-1 PRACE research supercomputer. Finisterrae III Supercomputer has 354 computing nodes that provide a total of 708 Intel Xeon Ice Lake 8352E processors, with 32 cores at 2.2 Ghz each. So there is a total of 22.656 processor cores. A total of 80 of this nodes include GPU acelerators (total of 128 Nvidia A100 + 16 Nvidia T4). All of this nodes together provide a peak computing power of 4 PetaFLOPS, 118 TB of main memory, and an High performance parallel file system lustre based. There is also a total 355 TB SSD NVMe storage that provide ultra high speed local data access at each node.
The computing nodes are connected through a low latency infiniband HDR100 network. CESGA storage facilities include an enterprise class 1,3 Petabyte NAS system which provides shared space to all of the computing servers, and an HSM system which includes a 4 Petabyte online disk space plus a tape library of 1.600 slots and 12 LTO9 + 4 LTO6 cartridges.
SPECIAL ACCESS CONDITIONS
No
CASE EXAMPLES
Simulations, HPDA, Scientific apllications
SUCCESS STORY
Nologin Consulting and Services is an spanish SME that has been using this service for providing IBI MFC Ocean Analisys and Simulation for Copernicus area. More info here
SERVICE CAN BE COMBINED WITH
Access to HPC Computing and Storage resources Access to Cloud computing resources Support of the desing and deployment of data infraestructures for analytics and visualization Support and/or execution of analysis of data from manufacturing process Support and/or implementation of feature selection techniques Design and development of AI models, their training and validation