Speeding Up Science Using GPUs
ESRL's Global Systems Division (GSD) develops next-generation weather models and has historically advanced high-performance computing for NOAA due to their need for larger, more powerful computers. Researchers at GSD began exploring GPUs a year ago and recently completed the parallelization of a large portion of the Non-hydrostatic Icosahedral Model (NIM) (the next generation Flow-following, finite-volume Icosahedral Model or FIM) and ran it on a GPU. The model ran twenty-five times faster on the GPU than the CPU. This is a significant and exciting result: it will allow NOAA scientists to continue developing weather and climate prediction models that use next-generation HPC systems to produce more timely and accurate forecasts.
Graphical Processor Units (GPUs) are considered by many to be the next frontier in High Performance Computing (HPC). CPU systems requiring over 200,000 cores are being built toward achieving over a petaFLOPS of computing necessary to run climate models and global weather models at cloud-resolving scales (3-4 km horizontal resolution) within three years. However, systems of this size are impractical for weather forecasting for many reasons including power and cooling requirements, infrastructure costs, system reliability, and power costs. Power bills alone account for over $1M in moderate-sized systems today; future CPU-based systems being proposed could require $100M annually if conventional CPUs are used.
GPUs are cheap, powerful processors designed for life-like video games and until recently used primarily by the gaming community. The HPC community's interest in GPUs was sparked when a high-level language called CUDA was released that allows programmers to write code for GPU processors. With CPU performance stalling, vendors have been forced to increase the number of cores per chip at the expense of increased power and cooling requirements.
By providing faster, cheaper calculating power and volume-handling storage of our supercomputers, our scientists can produce more accurate ocean, air quality, and environmental models that lead to a better understanding of our complicated Earth system.
Name: Mark W Govett