Computing resources

The IT resources in the Laboratoire de Biochimie Théorique have evolved considerably in recent years. This reflects the inherent increase in calculation demands of the laboratory and the renewal of the IT team, and follows particularly the awards of numerous European and national endowments including the ERC "Thermos", the ERC "ABIOS", the Labex "Dynamo", and several ANR grants.


Indeed, since 2011, the Laboratoire de Biochimie Théorique has acquired 3 high performance computing clusters (HPCCs):


  • Hades (2011-2012): HPCC composed of 29x bi Intel Xeon X5650 (Westmere platform) nodes totalising ~1.2TB of RAM (4GB per core) and up to ~3.7 CPU TFLOPS (FP64)

  • Lucifer (2013-2014): HPCC composed of 28x bi AMD Opteron 6380 (Abu Dhabi platform) nodes totalising 896GB of RAM (1GB per core) and up to ~9.2 CPU TFLOPS (FP64)

  • Baal (2015 till now): HPCC, which totals 1TB of RAM and can reach ~16 CPU TFLOPS (FP64) and be accelerated by 345 GPU TFLOPS (FP32), is composed by:


    • 1x bi Intel Xeon X5560 (Nehalem platform) node for test jobs only.

    • 2x bi Intel Xeon E5-2697v2 (Ivy Bridge platform) nodes for post-processing jobs

    • 3x bi Intel Xeon E5-2650v3 (Ivy Bridge platform) / bi Nvidia GeForce GTX-1080Ti nodes

    • 12x bi Intel Xeon E5-2698v4 (Broadwell platform) / bi Nvidia GeForce GTX-1080Ti nodes.

 


These computing resources are equipped with corresponding storage volumes appropriate to the required calculation intensity:


  • 408TB raw of computing storage volume

  • 336TB raw of archive volume

  • ~85TB of temporary computing space

Since our financial allocations are limited, and for better project control, we have fully designed and deployed these computing and storage ressources with open-source solutions by home-made way.


Our server room, located on the 3rd floor of the Institut de Biologie Physico-Chimique, has been completely redesigned to accommodate these HPCCs, our lab information system, and related services for other labs in the Institute.


The following schema provides a short introduction to our IT infrastructure and services:



For more information about our computing resources (for both info and use guide), you can access the dedicated wiki.