Supercomputing Challenges and Predictions
What will future supercomputers bring to the world? Personally, we hope that they will be used to finally design George Jetson’s flying car. But there are better experts out there making smarter predictions.
The future was, of course, a popular topic at SC12 in Salt Lake City last November. No flying cars, but increased energy efficiency, improved weather forecasting, a better understanding of the universe, and faster discovery of new drugs were all on the agenda.
IEEE has now put together a summary of supercomputing predictions and challenges made by a few of its members at SC12.
Rajeev Thakur, technical program chair of SC12 and deputy director of the Mathematics and Computer Science Division at Argonne National Laboratory, wins the prize for the most predictions. He foresees better batteries: materials science studies will enable people to create cheaper batteries with more capacity. Somebody has to replace the Energizer Bunny – and back up datacenters.
Thakur also believes cosmological simulations will answer questions about dark matter and dark energy, the geometry of the universe, and why the universe’s expansion rate is accelerating. (IBM’s next supercomputer will be named Einstein). Molecular simulations will create better drugs faster.
Energy was a popular theme. Bronis de Supinski, co-leader of the Advanced Simulation and Computing program’s Application Development Environment and Performance Team at Lawrence Livermore National Laboratory, favors the ability to better predict electricity demand on the grid. That means less wasted energy – and, perhaps, the ability to keep your computer from crashing in a power outage.
There were a few people predicting nuclear fission to solve the world’s energy problems (and, presumably, put the oil companies out of business). It’s just a matter of convincing people who still remember Three Mile Island.
A few had to put a damper on the rosy future, noting that there are still hurdles ahead. While de Supinski envisions better power grids, he also warns that the need for cheaper power and less dissipation en route will continue to present problems. That exascale computer you’re designing may have to wait a few more years before you plug it in. Thakur agrees with that one.
De Supinski also believes memory bandwidth and capacity will continue to fall behind computational power until applications are severely limited by these bottlenecks. Thinking deep thoughts isn’t much good if you can’t remember them.
And funding will, of course, be a problem, says Thakur. Perhaps, with luck, the Sequester will be over within a few years.
And, perhaps, at SC13 one of the supercomputers will itself be making the predictions.