SuperLU Users’ Guide∗ Xiaoye S. Li† Piyush Saok James W. Demmel‡

Source URL: crd-legacy.lbl.gov

Reproducible Tall-Skinny QR James Demmel and Hong Diep Nguyen ∗ University of California, Berkeley, Berkeley, USA

Source URL: bebop.cs.berkeley.edu

Corvette: Program Correctness, Verification and Testing for Exascale Koushik Sen and James Demmel (UC Berkeley) 1. Motivation q  High performance scientific computing"

Source URL: crd.lbl.gov

Write-Avoiding Algorithms Erin Carson James Demmel Laura Grigori Nicholas Knight Courant Institute of

Source URL: www.cs.huji.ac.il

Numerical Accuracy and Reproducibility at ExaScale James Demmel, Hong Diep Nguyen ParLab - EECS - UC Berkeley ARITH 21 April 7-10, 2013

Source URL: bebop.cs.berkeley.edu

CORVETTE: Program Correctness, Verification, and Testing for Exascale Koushik Sen (PI) James Demmel, University of California at Berkeley Costin Iancu, Lawrence Berkeley National Laboratory The goal of this project is to

Source URL: crd.lbl.gov

Optimization of Sparse Matrix-Vector Multiplication on Emerging Multicore Platforms Samuel Williams∗†, Leonid Oliker∗, Richard Vuduc‡ John Shalf∗, Katherine Yelick∗†, James Demmel† ∗ CRD/NERSC, Lawrence

Source URL: crd.lbl.gov

A Graph Expansion and Communication Costs of Fast Matrix Multiplication Grey Ballard, University of California at Berkeley James Demmel, University of California at Berkeley ¨ Berlin

Source URL: www.cs.huji.ac.il

• Mathematics
• Algebra
• Computational complexity theory
• Analysis of algorithms
• Multiplication
• Numerical linear algebra
• Matrix multiplication algorithm
• Matrix multiplication
• Strassen algorithm
• Fast Fourier transform
• Multiplication algorithm
• Algorithm

Graph Expansion and Communication Costs of Fast Matrix Multiplication Grey Ballard ∗ James Demmel †

Source URL: www.cs.huji.ac.il

• Graph theory
• Mathematics
• Combinatorial optimization
• Edsger W. Dijkstra
• Matrix multiplication algorithm
• NP-complete problems
• Matching
• Routing algorithms
• Computational complexity theory
• Connected component
• Dynamic programming
• Graph coloring

Fast Reproducible Floating-Point Summation James Demmel, Hong Diep Nguyen ParLab - EECS - UC Berkeley ARITH 21 April 7-10, 2013

Source URL: bebop.cs.berkeley.edu

• Computer arithmetic
• Computing
• Fixed-point arithmetic
• Multiplyaccumulate operation
• Precision
• Rounding
• FLOPS
• 128-bit
• Machine epsilon
• IEEE floating point