Home Research “Inexact computing” improves quality of simulations run on supercomputers

“Inexact computing” improves quality of simulations run on supercomputers


Simulations on supercomputers could receive a huge boost through “inexact computing” a new study led by Rice University researchers has demonstrated.

Using one of Isaac Newton’s numerical methods, researchers have shown that it is possible to dramatically improve resolution of weather and climate models with new ultra-efficient approaches to supercomputing. The latest study published in a paper on the preprint server ArXiv has its roots in the idea that was put forward by RUCCAM Director Krishna Palem in 2003.

Palem explains that there are calculations where accuracy to seven or eight decimal places is not needed and the results are perfectly fine even when they are accurate to just three or four decimals. Increased accuracy brings with it massive costs and ‘inexact computer’ is something that helps save energy by only paying for accuracy when it is actually needed. Palem adapted these approaches to climate and weather modeling, collaborated with Oxford University physicist and climate scientist Tim Palmer to show that inexact computing could potentially reduce by a factor of three the amount of energy needed to run weather models without compromising the quality of the forecast.

Researchers showed it is possible to leapfrog from one part of a computation to the next and reinvest the energy saved from inexact computations at each new leap to increase the quality of the final answer while retaining the same energy budget. Palem likened the new approach to calculating answers in a relay of sprints rather than in a marathon.

“By cutting precision and handing off the saved energy, we achieve significant quality improvements,” said Palem, Rice’s Kenneth and Audrey Kennedy Professor of Computer Science. “This model allows us to change the way computational energy resources are utilized in supercomputers to dramatically improve solutions within a fixed energy budget.”

The research team took advantage of one of the most commonly used tools of numerical analysis, a method known as Newton-Raphson that was created in the 1600s by Isaac Newton and Joseph Raphson. In supercomputing, the method is used to allow high-performance computers to find successively better approximations to complex mathematical functions.

The researchers demonstrated that the solution’s quality could be improved by more than three orders of magnitude for a fixed energy cost when an inexact approach to calculation was used rather than a traditional high-precision approach.

Palem said, “A specific goal is to encourage the application of this approach as a way to advance the quality of weather and climate modeling by improving model resolution.”

He said RUCCAM is working with Oxford’s Palmer and others to explore possible ways to improve the resolution of the OpenIFS model that was developed by the European Center of Medium Range Weather Forecasting.

RUCCAM brings together researchers from Rice and universities around the world to explore solutions to the physical and energy limitations currently restricting the continued expansion of computing capacity needed to solve emerging workload problems. The researchers intend to change the manner in which computational resources are utilized, even at the margins of stability and accuracy, to increase the efficiency at which answers can be calculated.