High Performance Computing Environment

P. G. Senapathy Centre for Computing Resources
Login






HIGH PERFORMANCE COMPUTING ENVIRONMENT FACILITY


A Resource for Research and Discovery

The High Performance Computing Environment (HPCE) is set up to cater to the ever increasing demand, for supercomputing facilities, of researchers at IIT Madras. Material science and engineering, atmospheric and ocean modeling, aerospace engineering, modeling social, ecological and physical networks, design of large structures and VLSI, understanding flows and combustion, spectroscopy and molecular modeling are some of the active research areas in IIT Madras, where such large scale computing will be used.

High Performance Computing (HPC), (also known as: Research Computing or Scientific Computing) is for those computing tasks which for some reason or another just won't run on a desktop computer. The task may be to sort through a huge amount of data, solve a difficult numerical problem, or just repeat the same calculation over and over for different situations. There are basically three types of High Performance computing architectures available, SMP (Symmetrical Multi Processor computing), Cluster computing and Grid computing. Each of these architectures is designed to handle different types of computing problems based on what is needed.

a.SMP system is a computer that contains multiple processors, shares the same memory and is managed by one operating system. These SMP systems are normally used for computing jobs that require large amount of memory.
b.Cluster Computing on the other hand is a group of systems (nodes) linked together through some sort of interconnect and are used for parallel processing for scientific applications. Typical users of Clusters will break their compute job into pieces, distribute the pieces across many nodes on the cluster and accumulate the results in the end.
c.Grid Computing focuses on maximizing the use of processor resources of various computers (possibly throughout a large geographical area). Each of these computers is running a software client that allows access to system and the processor to be used when the computer is powered on and no one is using it. This is a very efficient in that grid software can take advantage of computer processors of a variety of platforms and system configurations.

HPC Computing Resources
The High Performance Computing environment consists of high-end systems used for executing complex number crunching applications for research it has two such machines and they are,
  • AQUA Super Cluster
  • GNR Super Cluster
The machines in the above list belong to cluster computing category.

AQUA Super Cluster
System Configuration
  • 260 CPU only Compute Nodes
  • 12 GPU Nodes
  • 2 Master Nodes
  • 1 Login Node
    Total Compute Power 700 TFlops
    HPE Apollo XL170rGen10 Servers. Populated with Dual Intel Xeon Gold 6248 20-core, 2.5 GHz processors (Standard Compute Nodes)
    HPE Apollo XL190rGen10 Servers. Populated with Dual Intel Xeon Gold 6142 16-core, 2.6 GHz processors (GPU Nodes)
    A total of 192 GB RAM per node with storage of 1 x 2 TB SATA disk
    Single port Mellanox HDR100 HCA (Primary Interconnect)
    2 x NVidia V100 32GB GPUs – PCIe

    Storage Configuration
  • 1 PetaByte PFS (HPE Lustre Storage) with minimum 25 GB/s write performance.
  • 200 TB NAS Storage

    GNR Super Cluster
    This cluster is named after the scientist Prof. G N Ramachandran
    • 1 Head Node on Super micro servers with Dual Processors, Eight-Core Intel Xeon Ivy bridge E5-2650v2 series processors with 4 X 8GB RAM and 500 GB of SATA Hard disk.
    • 16 compute nodes based on super micro server with Dual processor, Eight-core Intel Xeon Ivy Bridge E5-2650v2 series Processors with 4 X 8 GB RAM and 500 GB of SATA Hard disk in each node.
    • 14TB of shared storage

    Documents:
    HPCE PRESENTATION