NATIONAL TAIWAN UNIVERSITY CHOOSES GPUS TO UNLOCK MYSTERIES OF THE UNIVERSE

by Andrew Humber

GPU-powered supercomputers are cropping up everywhere, tackling some of the biggest scientific puzzles today. Now they’re even cracking the mysteries of the universe. A research team at National Taiwan University (NTU) announced that it is using NVIDIA Tesla processors to power its research into sub-atomic particles and their role in the origins of the universe. Check out the slide deck that NTU presented at this year’s GTC.

NTU is using Taiwan’s first GPU-based supercomputer – a 128-GPU cluster using 16 NVIDIA Tesla S1070 1U systems and 64 Tesla C1060 processors – for everything from research on hadrons at the sub-atomic scale to cosmology at the astronomical scale.

Large-scale computations for quantum physics are so compute-intensive, they’ve typically been run on conventional supercomputers, like the IBM BlueGene/L. But we’re seeing the game change in research as more labs and universities realize they can use GPUs to get the same performance of traditional supercomputers at a fraction of the price. The NTU GPU cluster, for example, is delivering the same computational power of a BlueGene/L for 1% of the price and 10% of the power consumption. For big-brain scientists, this particular calculation is a no-brainer.