New FEDER-CPER IT2MP Cluster for distributed computing on large graphs
Thanks to the funding of CPER IT2MP, including FEDER contribution, a new cluster called Grappe has been installed in the Grid5K environment on 16 October 2020. Grappe has been configured to allow efficient computation on large distributed graphs. In brief, it is composed of 16 nodes (PowerEdge R640), each carrying 2 CPUs with 20 cores, 96GB memory, 480GB SDD and 8TB HDD. We are grateful to the above-mentioned financial support and to the engineers of the Grid’5000 technical team for the successful installation of Grappe.
New storage facility for the MBI-DS4H platform thanks to FEDER-CPER IT2MP funding
On 4 september 2020, a new machine has been installed on the platform to facilitate the secured storage and access to data files. The machine (PowerEdge R740XD2) is called mbi-storage and is equipped with 12 Hard Drives Hot Plug 12To NL-SAS 12Gbit/s 512e 7200tr/min 3,5″. We acknowledge the CPER IT2MP funding, including the FEDER contribution and the help of our engineers Patrice Ringot and Jonathan Alcuta for taking care of the configuration and installation of this server.
- Two clusters (MBI- and SB- servers composed of 12 and 28 bi-quad processors connected in Infiniband, and extended by 4 Teslas each for GPU computing (see complete description below)
- Storage servers (about 40 To)
- Application servers
- High-performance 3D visualisation station
- Web server
- High-Performance computing is also possible on Grid’5000 facilities at the LORIA
The MBI server (2012) is composed of
- A master node, with 2 processors Xeon X5680 3.33G 6 cores, 300Go storage,
- 4 NVIDIA FERMI C2070 GPU 448 CUDA cores,
leading to a total of 12 CPUs and 1792 CUDA cores.
The SB-server (2016) is composed of several units, interconnected with Mellanox connectX-3:
- a storage server (PowerEdge R730xd), Processor Intel Xeon ES-2680 v4 2.4 GHz 14 cores / 28 Threads, RAM 24×32 Go = 768 Go, Storage 4×10 To = 40 To ;
- Two HPC servers (PowerEdge R730) composed of
- 2 processors Intel Xeon ES-2680 v4 2.4 GHz 14 CPUs / 28 Threads,
- 2 Teslas GPU NVIDIA K40M with 2880 CUDA cores,
leading to a total of 56 CPUs (112 threads) with 8×16 = 96 Go RAM, and 11520 CUDA cores with 4×12 = 48 Go RAM.
Since 2011 the MBI platform is part of the North-East (NE) node of the French RENABI Network of Bioinformatics Platforms, together with the Lille, Strasbourg, and Reims bioinformatics platforms, under the coordination of Dr. Maude Pupin in Lille.
Since 2013 the MBI platform and the NE node of the French RENABI network are members of the French Institute of Bioinformatics.
The MBI platform’s manager is Dr. Marie-Dominique Devignes, member of the CAPSID team at the Loria since january 2015. The platform operating committee is composed of Loria main users of the platform.
Most external users of the platform are partners involved in multi-disciplinary projects with Loria computer scientists. All external users need to request an account for using the computing resources of the MBI platform.