Overview Of Sankhya Clusters:

Sankhya Clusters consists of one Master Node and three or more Compute Nodes. The Clusters are available in 4 Nodes and are scalable to 8 Nodes, 16 Nodes, 24 Nodes, 32 Nodes, 48 Nodes, 64 Nodes and 128 Nodes and above. We also offer "customized clusters", based on the customer's hardware specifications and budget

Why Sankhya Clusters?

As computational requirements grow by the minute, customers require more cost-effective clusters for their needs. Sankhya clusters are affordable, Reliable and Scalable (RAS), has a Lower `Cost Of Ownership (COO)'. These clusters also give an added benefit of Better `Return On Investments (ROI)'. The PRICE Vs PERFORMANCE of Sankhya is much better than most other traditional and commercial Supercomputers. Our customers will agree with us when we say that Sankhya HPC clusters are easy to deploy, maintain and service.

Sankhya Composes Of

Chassis

Sankhya Cluster uses lightweight, thermally optimized, vertical rack mountable chassis specially designed by TAURUS, which fit into our Racks. We make both horizontal and vertical rack mountable chassis.

Rack

TAURUS has specially designed Aluminum Racks of 24U/36U/42U Height, 800/1000 mm Depth with/without cooling fans. We custom build in various dimensions to suit specific clusters. The issues concerning airflow, heat dissipation and thermal problems have been effectively handled.

Our Speciality - Vertical rack mounting

TAURUS specializes in `Vertical Rack Mounting', which most of the scientists believe is the best way to mount Servers inside a Rack. Our customized chassis are manufactured and mounted inside the racks on `rails'.

Hardware Speciality

The chassis are mounted in such a way that there is a smooth passage for air-flow from the bottom of the chassis to the top. There are 4 such air passages for a 4 X 4 matrix. We use 4 fans on the bottom of the rack with Air Filters to blow cold air and 4 fans on the top of the rack to suck out the hot air.Since, the air within the chassis and rack is circulated regularly, there wont be thermal locks inside the rack, which is the prominent design feature of Sankhya HPC Clusters. This feature is of greater importance in the coming days than ever before, as the clock speed of CPUs are increasing and the silicon thickness in the CPU are becoming smaller which results in lot of heat being generated. Typically, in a 42U Rack, we mount up to 16 Compute Nodes as a matrix of 4 X 4. Also save in cutting down cost of cooling.

Hardware Testing

All the hardware like Motherboard, Memory, Hard Disk, etc. undergo `rigorous and standard hardware tests before integration. The cluster is also tested for inter-connect, OS stability, performance, etc.

Benchmarking

We have been able to get very good performance and scaling on our clusters. We usually benchmark our clusters with standard tests like HP Linpack or other proprietary codes from our customers.

Tools

We offer freely available and useful cluster monitoring tools. We also provide C / C++ / Fortran Compilers from Intel (Demo Version) or full version on payment.