A progressive migration of BiRD users to the regional mesocentre GLiCID is currently underway.
The reasons for this migration are the pooling of human and material resources on a regional scale, in collaboration with the Universities of Nantes, Angers, Le Mans and the Ecole Centrale de Nantes.
This pooling will enable us to offer you :
-
access to hours of CPU and GPU calculations for sequential and parallel computing (16,300 cores of all types, 100TB of RAM, 80 GPUs)
-
access to high-performance, volumetric data storage (1.5Pb of fast storage and 4Pb of CEPH warm storage)
The differences with the BiRD infrastructure services as you know them are as follows:
-
change of scheduler: SGE -> SLURM for better management of computing resources
-
access to higher performance computing nodes (latest generation CPUs): AMD Genoa and Intel Sapphire rappids
-
access to GPU nodes (A100-80GB)
-
access to BigMem nodes
-
access to fast SSD storage
Some of you have already taken the plunge, others not yet. Initial feedback has been very positive, particularly in terms of node resource management, GPU access and clearer data management.
