by Roy Kim

What if every graduate student, researcher and professor could have access to a supercomputer, right in their own university? Can you imagine what kind of scientific breakthroughs they could make with this level of computing power?

The problem with traditional supercomputers is that they are large and expensive, and so accessible to only a select few scientists. Today, however, there is an alternative to high-performance computers with large footprints and lofty price tags.

HP ProLiant SL390 G7 server with NVIDIA Tesla M2070 GPUs

HP has just launched the GPU Starter Kit, which makes it easier and more affordable than ever to put the power of a supercomputer in the hands of researchers everywhere. The Kit is a fully preconfigured cluster of HP servers accelerated by NVIDIA Tesla GPUs – the computational engine under the hoods of the world’s fastest supercomputers, like the Tianhe-1A in China and Tsubame 2.0 in Japan.

Better yet, at half the typical list price, the GPU Starter Kit is an attractive option for a wide variety of researchers and scientists. But, don’t think that “mini” systems like these are small on performance.

The fact is, amazing science is mushrooming all over the world on smaller installations of GPU-powered servers:

  • At the University of Delaware, researchers are accelerating discoveries in human pathological conditions. With GPUs, their research is done 10x faster.
  • Researchers at the University of Padua, Italy, are using a 16 GPU cluster to better understand the beginning of the universe.
  • In a joint collaboration between University of North Carolina Charlotte and Georgia Tech, researchers are looking to GPUs to better understand protein-DNA interactions for future drug design. With 32 GPUs, they are able to achieve higher performance than with 1,200 CPUs alone.

Nothing inspires us more than empowering scientists to do amazing things. We believe one way to equip the thousands of university researchers with the tools they need is to start small. That’s where the GPU Starter Kit comes in.

So what is the GPU Starter Kit from HP? It’s a pre-configured cluster of eight HP Proliant SL390 servers with 24 NVIDIA Tesla M2070 GPUs. It is packaged to help researchers hit the ground running with an easy out-of-the-box experience. More details on the configuration are available as a direct PDF download here.

Many researchers rely on popular scientific applications that are already accelerated with GPUs. But for those who own proprietary applications, the kit includes a broad set of development tools at special discount prices.

With this offering, HP and NVIDIA are putting mini-supercomputers right in the researchers’ offices, empowering them to do great things.  Ultimately this will mean even more breakthrough science with GPUs.

You can email us at hpstarterkit @ nvidia.com if you need more info.