Sign In

Communications of the ACM

ACM TechNews

Supercomputing Under a New Lens: A Sandia-Developed Benchmark Re-Ranks Top Computers


Reimagining the Top500 ranking using the High-Performance Conjugate Gradients benchmarking program.

The Top500 ranking of supercomputers would look different when using the High-Performance Conjugate Gradients benchmarking program, rather than LINPACK.

Credit: Sandia National Laboratories

Sandia National Laboratories' High-Performance Conjugate Gradients (HPCG) benchmarking program is gaining prominence as a tool for ranking supercomputer performance.

"The LINPACK program performs compute-rich algorithms on dense data structures to identify the theoretical maximum speed of a supercomputer," says Sandia's Mike Heroux. "Today's applications often use sparse data structures, and computations are leaner."

To better reflect the practical elements of current supercomputing application programs, Heroux devised HPCG's preconditioned iterative technique for solving systems containing billions of linear equations and billions of unknowns. "A preconditioner makes the iterative method converge more quickly, so a multigrid preconditioner is applied to the method at each iteration," Heroux notes.

Major supercomputer vendors are writing versions of HPCG's program that best suit their platforms.

From Sandia National Laboratories
View Full Article

 

Abstracts Copyright © 2018 Information Inc., Bethesda, Maryland, USA


 

No entries found