Harness the Power of Supercomputing Without Spending a Fortune

Summers are coming and the heat is just moderate enough to prevent the ice cream on the plate from melti ng too rapidly. However, there is more to the speed at which ice cream melts than heat. Scienti sts worldwide are using super computers to understand the fl uid dynamics of various liquids and semi- liquids so that we get bett er quality commercial products (and ice cream being one of them!).

GPUs & science of smooth ice cream

Simulati ons based around fl uid dynamics off er a powerful way to study, predict and ulti mately improve the behavior of "soft matt er" - materials such as paints, engine lubricants and even that tub of ice cream in your freezer. Over the last 12 years, a collaborati on between the University of Edinburgh's Soft Condensed Matt er Physics group and the Edinburgh Parallel Computer Centre (EPCC) has developed simulati ons of soft matt er systems using the latti ce Boltzmann method and the parallel comput- ing code "Ludwig" to accurately capture the Physics of systems such as mixtures, suspensions and liquid crystals. Understanding and controlling the phase separati on of liquid mixtures, for example, prevents the formati on of those ice cream-compromising ice crystals and improves the shelf-life of frozen desserts - one of the many practi cal applicati ons of the research. Now, researchers are exploiti ng the Cray XK6 supercomputer to accelerate their research. However, the challenge was to maintain good scalability when halo data must be exchanged be- tween many GPUs. To make sure they obtained the best performance scaling, the Ludwig developers used NVIDIA CUDA streams to overlap data transfers between the CPU and GPU. A crucial factor in obtaining accurate results is the size of the physical system that we can simulate," says lead Ludwig author Kevin Stratf ord (EPCC). "To model com- plex problems in large systems, we need effi cient, scalable applicati on performance over large numbers of nodes." And while companies like Cray and EPCC work towards realizing the exascale, it's good to know that one of the results is smoother ice cream.

High Performance Computing in India

Dr Pradeep Sinha, Senior Director, High Performance Computi ng (HPC), C-DAC, had been bestowed with the recogniti on of IEEE Fellow by the Insti tute of Electrical and Electronics Engineers (IEEE), New Jersey, USA, in 2012. research in their respecti ve domains that matches In 2011, C-DAC had bagged the Internati onal Data Corporati on (IDC) HPC Innovati on Excellence Award. The IDC HPC Innovati on Excellence Award recognizes notewor- thy achievements by users of High Performance Computi ng (HPC) technologies in Scienti fi c, Research, Academia and Government. The program's main goals are to showcase return on investment (ROI) and scienti fi c success stories involving HPC; to help other users bett er understand the benefi ts of adopti ng HPC.

How can you make use of HPC in your organization?

Freely available open-source soluti ons such as OSCAR (Open Source Cluster Applicati on Resources, htt p:// ld2.in/4gd ) can be used by organizati ons without facing a steep learning curve. No familiarity with *nix environments is necessary in order to make use of OSCAR. Given OSCAR's low system requirements (both for the server node in a cluster as well as for clients), organizati ons can put their existi ng hardware to task more effi ciently using OSCAR for compute- intensive jobs compared to using each node in an autonomous manner.