Home > About CRL > CRL Overview

CRL Overview

Computational Research Laboratories (CRL) a fully owned subsidiary of Tata Sons Ltd engaged in cutting edge research and development in the area of high performance computing. CRL’s interests include computer system architecture, parallel system design, parallel systems software, hardware design and interconnection networks, parallel scientific, engineering, digital media, communication and business analytics software libraries and applications. Aiming this goal and to be among the top research, technology and business leaders in the world in high performance computing CRL has decided to build a 100 plus teraflop sustained performance system based on parallel processing architectures.

With its core objective CRL has built the fastest supercomputer in Asia, and the 4th fastest supercomputer in the world. CRL's supercomputing facility hosts the ‘EKA’ supercomputer with 170 TeraFLOPs peak performance, and 120 TeraFLOPs sustained LINPACK performance built using dense data center layout and novel network routing and parallel processing library technologies. Built in just 6 weeks, EKA has 1800 nodes having dual-socket quad-core Clovertown CPUs from Intel. The nodes are connected using Infiniband 4X DDR switches. The system has high performance Storage Area Network using Infiniband with Lustre file system.

The facility aims to provide easy-to-use HPC development environment and a 24X7 application services, with a user friendly problem solving environment. The facility will provide time cycles on EKA with value added services combined with application domain expertise. A range of services from simple "CPU cycles for application development" to "hosted application services" for various domains of client interest are offered.

CRL also provides HPC system integration, HPC facility design, High Performance Storage design and HPC consultancy services. Future plans also include application porting services using multi core/multithread optimizations and optimal parallel processing techniques

Whitepapers

 
 
 
 
 
 

  © 2008 Computational Research Laboratories

Site Map | Contact us