SAHA INSTITUTE OF NUCLEAR PHYSICS
Department of Atomic Energy, Govt. of India
Webmail Member Search A+ A A-
Log in
tcmp

Facilities

Configuration of TCMP Computer Cluster - 1, Meghnad

Master Node

Hardware : IBM x3850M5 with 4 x Intel Xeon CPU E7- 4807, 40GB RAM, 3x146GB HDD
Operating System : Centos 6.2 – Linux Kernel 2.6.32-220.13.1.el6.x86_64 SMP
Cluster Suite : Rocks 6.0 with Sun Grid Engine (SGE) Tools, Ganglia etc.
Compilers : Gnu Compilers (gcc 4.4.6, gfortran) , Intel Compilers (ifort 12.0.0, icc 12.0.4)
Libraries : lapack, blas, Gnu Scientific Library (gsl)

Compute Node

Hardware : 14 x IBM Blade Server with 2 x Intel Xeon CPU X5670 6C/12T @ 2.93GHz, 48GB RAM, 146GB HDD hosted in IBM Blade Center H 

Size of Cluster
Total Number of Compute Nodes : 14 x (2X6X2) nodes = 336 nodes [ Hyperthreading Enabled ]

FLOPS
Theoretical Flops calculation : 4 flops/Hz X 2.93Ghz/core X (6 core / Processor X 2 Processors / Blade X 14 Blades) cores = 1968.96 Gigaflops ~ 1.969 TFlops

Documents
Commands to run jobs using cluster are available at here

Configuration of TCMP Computer Cluster - 2

Master Node

Hardware : IBM Power 7 8231-E1C with 8 node 3 GHz  , 8 GB RAM,  300 GB HDD
Operating System : SuSE Linux ES 11 – Linux Kernel 3.0.58-0.6.6 PPC64 SMP
Cluster Suite : Torque Power PC 64
Compilers : XL Fortran and C Compilers
Libraries : Scientific Library (essl)

Compute Node

Hardware : 2 x IBM Power 7 8236-E8C with 128 x IBM Power7 3.6GHz , 128GB RAM

Size of Cluster
Total Number of Compute Nodes : 2 x 128 nodes = 256 nodes

FLOPS
Theoretical Flops calculation : 8 flops/Hz * 3.6 Ghz/core * 256 cores = 7372.8 Gigaflops~ 7.3 TFlops



 

Last Updated on Thursday, 10 August 2017 15:49
Copyright © 2017 Saha Institute of Nuclear Physics. All Rights Reserved. Designed & Developed by UCDS