The Challenge of Science and Medicine

Whether it's geological survey's, computational fluid dynamics (CFD), molecular dynamics, medical imaging (e.g. MRI), physics simulations or protein folding a personal supercomputer can rapidly accelerate research and development work. From eight to over 1000 cores are available to be used for crunching through calculations in parallel dramatically reducing time to complete. Typically large in memory data sets can slow processes down by requiring page faulting out to virtual memory, or memory fragmentation. With over 100GB of available memory its possible to retain whole arrays, datasets, databases or processes in memory for rapid access and analysis. The new platform offers opportunities to transform the way research is carried out and speed the time to market of the end product.

Practical Applications

The Lattice Boltzmann method (LBM) - Is used for modelling two (shallow) and now three-dimensional water flow fields coupled to mass transport. The LBM is one of a few methods used to solve various fluid dynamics phenomena. LBM equations have wide applications in ocean and hydraulic engineering which can benefit from the advantages of the method. Many-fold performance advantages offered by the Personal Supercomputer or High Performance Computer (HPC) have opened up new applications that were previously technically impractical to attempt due to the sheer volume of calculations required.

From lethal hurricanes to global climate change - Weather is one of the biggest uncontrollable factors affecting our daily lives and in some geographies peoples very survival. Early predictions of weather events are becoming faster and more accurate, giving people more time to prepare, thus saving lives and livelihoods. Critical to the ability to create timely and accurate models is the volume of calculations and simulations you are able to run and the amount of data you are able to process. Personal Supercomputers and High Performance Computing (HPC) have extended the technical feasibility and capability of models radically. At the National Centre for Atmospheric Research (NCAR) a team of scientists has developed sophisticated forecasting models for immediate, long-term and short-term weather conditions. The Weather Research & Forecasting Model (WRF) is the most widely used model in the world, with users including the National Weather Service, the Air Force Weather Agency, foreign weather services and commercial weather forecasting companies. NCAR's climate and weather models are moving from Terascale (1 trillion FLOPS) to Petascale class applications, outgrowing conventional computing clusters and reaching a tipping point where adding more CPUs is no longer effective for improving speed. The problem has been particularly acute with applications that involve a real-time component or other time-to-solution constraints.

NCAR technical staff collaborated with researchers at the University of Colorado and turned to the NVIDIA Personal Supercomputer GPU Computing solutions. There was a ten fold improvement in performance for Microphysics a computationally intensive component of WRF. Although Microphysics made up a small fraction of the model source code, accelerating it on the Supercomputer platform resulted in a 20 percent speed improvement for the overall model. Demonstrating that that its not the code volume that's important its the execution density over a given fundamentally important algorithm or calculation that will deliver the benefits. "This result is extremely encouraging and timely, coming at a time when we are running out of gas for time-critical forecasts on conventional clusters," says NCAR's John Michalakes, lead software developer for WRF. "We aim to cut the time for a forecast by at least a factor of two as we incorporate NVIDIA's GPU computing technology into more of WRF. I expect the affect of accelerators in weather and climate modelling will be transformative."

Cryo PC and Nehalem extends the computational power envelope

For traditional main processor work the Nehalem platform offers up to 24 simultaneous processing threads of execution (half of those utilising HyperThreading to provide virtual processing cores). The available over performance from Cryo Boost lifts the processor core speed from an Intel factory standard of around 3GHz to as much as 5GHz or more. In addition Cryo Boost also increases bus bandwidths so that memory, PCI and Quick Path Interconnect buses operate at around 50% over standard. Memory speeds also operate with similar over performance reaching speeds in excess of 2.3GHz, up to 1GHz over the factory standard.

Grow and Upgrade with the platform

With the Cryo PC's modular, standardised and component based design support, maintenance and upgrades are the same as a standard PC. While you are leveraging a personal desktop supercomputer asset you are not required to pay the traditional acquisition, maintenance and replacement costs of an industrial supercomputer. It requires no special facilities or skills to maintain and can even be deployed with a corporate standard build, with a few minor modifications. The standard machine can be built with four to eight main cores. By leveraging GPU processing using CUDA up to an additional 240 mid-speed processing cores can be added per GPU (Tesla) card. As new GPU's are released new cards can be added and/or replaced to provide yet more processing cores.