Resources
Technical and scientific high performance supercomputing is becoming more and more important. Realistic combustion computation using complex geometries, chemical kinetics, aircrafts aerodynamics, climate variability ... for each of these CERFACS' research domains high performance conputing is a prerequisite to stay competitive.
Indoors CERFACS' computing servers deliver about 70 Tflop/s :
BULL Intel SandyBridge cluster : BULL B510 - 41 peak Tflop/s
HP AMD MagnyCours cluster : HP C7000 - 20.3 peak Tflop/s
IBM Intel Nehalem cluster : iDataPlex - 7 Tflop/s
Each of these supercomputers access to their local parallel filesystem (based on GPFS technologies for IBM's systems or Luster for HP server).
About 200 TO of second level storage is distributed through NFS by an EMC VNX 5700
Scalar Computing : BULL Novascale (CEA CCRT & GENCI, PRACE), SGI ICE (CINES) & IBM Power7 (ECMWF), IBM BG/P (Argonne National Laboratory and Idris).
These resources allow CERFACS' to port and test new technologies closely to our industrials partners.
Through CERFACS' Partnerships additional resources are available outdoors :
Vector Computing : NEC SX8R & NEC SX9 (Meteo-France).Scalar Computing : BULL Novascale (CEA CCRT & GENCI, PRACE), SGI ICE (CINES) & IBM Power7 (ECMWF), IBM BG/P (Argonne National Laboratory and Idris).
Thanks to suppliers we access to additional technology watch resources :
Westmere server and NVidia TESLA GPU card
INTEL pre GA server's loan.
These resources allow CERFACS' to port and test new technologies closely to our industrials partners.
