Several countries are ramping up their so-called exascale programs, which are aimed at building computers 1,000 times more powerful than today’s most powerful machines (and able to do one quintillion floating ops per second). These large-scale projects have halo effects, inspiring parallel development in processor, storage, and networking technologies. So what’s America’s problem? InfoWorld has a list:
1: The U.S. doesn’t have an exascale plan.
Europe, China, and Japan have plans and money. We don’t. The U.S. is funding IBM’s planned 20-petaflop computer for Lawrence Livermore National Laboratory, but there’s no federal budget for exascale development and there likely won’t be, given the economic mess we’re in.
2: We assume we’ll win anyway.
We don’t have the political will or financial resources to build huge projects like the supercollider once planned for Texas, and we assume others don’t either. And yet, look at the Large Hadron Collider in Europe, which has attracted brain power, development, and excitement.
3: There’s no clear path forward.
The U.S. has outlined some requirements for an exascale system, saying it must be ready by 2019-2020 and can’t use more than 20 megawatts of power, which may be simply impossible to achieve.
4: If we mess up now, we’ll be way behind in the next round
After exascale systems will come zetascale systems, which will be another 1,000 times more powerful. Such a system may arrive around 2030, but the way things are going, it’s highly unlikely that it will be American-made.
5: We don’t understand why all this matters.
President Obama has mentioned exascale computing, but it was mostly lip service. A clear case must be made for supercomputing that explains how it addresses medical research, climate change, design modeling, earthquake prediction, and more. It’s not just the computer itself. It’s everything the computer can do.