When did you last look up at the night sky and really look at the stars?
Astronomers at the U.K.’s Durham University are exploring ways to understand the early universe, and the one seen from our backyards, through simulations performed using GPU-powered high performance computing.
World-renowned for teaching and research, Durham is using HPC to deepen our understanding of astronomy. As one of our newest CUDA Research Centers, it will use GPUs to tackle the astronomical computing challenges associated with the work.
Durham joins 40 other institutions added during the past quarter to our roster of CUDA Research and Teaching Centers, bringing the total to 431 in 55-plus countries. The goal: prepare researchers, engineers and computer scientists for groundbreaking work across an array of fields using GPU accelerators.
Star Power: Durham Using GPUs to Speed Simulations
Durham has embraced NVIDIA Tesla and GeForce technology within several departments, notably physics, whose astronomers make observations of the universe using some of the world’s largest telescopes and which hosts both the Institute for Computational Cosmology and the Centre for Advanced Instrumentation.
Durham’s researchers are studying properties of the Milky Way and simulating the dark matter known to exist due to its gravitational effects, but that we can’t see.
To ensure that its computer simulations are accurate, they compare them to actual observations. However, images collected by ground-based telescopes are distorted by passage through Earth’s atmosphere – which causes the “twinkling” of stars. Researchers correct the effect using sophisticated mirrors, super-high-speed cameras and robust computer power, which tests show can be amply provided for using GPUs.
But even with hundreds of computers clustered together, HPC simulations can take weeks. Ever larger simulations—and comparisons to observational data from ever larger telescopes—require a significant jump in computational power.
And here is where the power of GPUs can shine as brightly as the stars being scrutinized.
Durham is modifying its HPC software to make use of GPUs, which are expected to provide a substantial acceleration so that the simulation algorithms can be run faster and in more detail.
New CUDA Research Centers
CUDA Research Centers embrace GPU computing across research fields. They get access to events with key researchers and academics, an NVIDIA technical liaison and specialized training sessions. Our newest ones, in addition to Durham, include:
- Hamburg University of Technology (Germany)
- Harbin Institute of Technology (China)
- Indian Institute of Technology, Roorkee (India)
- Khmelnytskyi National University (Ukraine)
- Michigan Technological University (U.S.)
- Pontificia Universidad Católica del Perú (Peru)
- Purdue University (U.S.)
- Riga Technical University (Latvia)
- Singapore Management University (Singapore)
- SURFsara (Netherlands)
- Ufa State Aviation Technical University (Russia)
- Universität Passau (Germany)
- University of Las Palmas de Gran Canaria (Spain)
- Wright State University (U.S.)
New CUDA Teaching Centers
CUDA Teaching Centers ready tens of thousands of students graduating each year to take advantage of the power of GPUs. They get teaching kits, textbooks and software licenses, plus CUDA architecture-enabled GPUs for teaching labs and discounts for extra hardware.
- 3DMX University of Advanced Technologies (Mexico)
- College of William and Mary (U.S.)
- Colorado School of Mines (U.S.)
- D. Y. Patil College of Engineering (India)
- Federal University of Western Pará (Brazil)
- Harran University (Turkey)
- Jiangsu University of Science & Technology (China)
- Malaviya National Institute of Technology (India)
- Minia University (Egypt)
- Óbuda University (Hungary)
- Punjab University College of Information Technology (Pakistan)
- R.H. Sapat College of Engineering, Management Studies & Research (India)
- Riga Technical University (Latvia)
- Sardar Patel Institute of Technology (India)
- Shandong University (China)
- Shri Dharamasthala Manjunatheshwara College of Engineering and Technology (India)
- Silver Oak College of Engineering & Technology (India)
- Sinhgad College of Engineering (India)
- Texas Tech (U.S.)
- Universidad Rey Juan Carlos (Spain)
- University of Central Missouri (U.S.)
- University of Malta (Malta)
- University of Pennsylvania (U.S.)
- University of São Paulo (Brazil)
- University of Valladolid (Spain)
- Warsaw University of Technology (Poland)
For more information on NVIDIA research activities and these programs, visit the NVIDIA Academic Programs website.
Headline image courtesy Andrew Reeves – CfAI