This workshop will be offered virtually. The in-person meeting has been cancelled due to the COVID-19 outbreak. A schedule for virtual talks will be posted soon. Accepted participants will be notified how to access the virtual presentations. All other interested parties can view the talks via live-stream.
Organizing Committee
Abstract

From its introduction in the 1980s, the IEEE-754 standard for floating-point arithmetic has ably served a wide range of scientists and engineers. Even today, the vast majority of numerical computations employ either IEEE single or IEEE double, typically one or the other exclusively in a single application. However, recent developments have exhibited the need for a broader range of precision levels, and a varying level of precision within a single application. There are clear performance advantages to a variable precision framework: faster processing, better cache utilization, lower memory usage, and lower long-term data storage. But effective usage of variable precision requires a more sophisticated mathematical framework, together with corresponding software tools and diagnostic facilities.

At the low end, the explosive rise of graphics, artificial intelligence, and machine learning has underscored the utility of reduced precision levels. Accordingly, an IEEE 16-bit "half" precision standard has been specified, with five exponent bits and ten mantissa bits. Many in the machine learning community are using the "bfloat16" format, which has eight exponent bits and seven mantissa bits. Hardware such as NVIDIA's tensor core units can take advantage of these formats to significantly increase processing rates.

Image for "VIRTUAL ONLY: Variable Precision in Mathematical and Scientific Computing"

Confirmed Speakers & Participants

Talks will be presented virtually or in-person as indicated in the schedule below.

  • Speaker
  • Poster Presenter
  • Attendee
  • Virtual Attendee

Workshop Schedule

Thursday, May 7, 2020
TimeEventLocationMaterials
9:50 - 10:00am EDTWelcome - ICERM Director  
10:00 - 10:45am EDTVariable precision computing- Applications and challenges - David H. Bailey, University of California, Davis 
11:00 - 11:15am EDTCoffee/Tea Break  
11:15 - 12:00pm EDTHow Slow is Quadruple Precision - Paul Zimmermann, INRIA, Nancy, France 
12:15 - 1:30pm EDTBreak for Lunch / Free Time  
1:30 - 2:15pm EDTDynamic Analyses for Floating-Point Precision Tuning - Cindy Rubio Gonzalez, University of California, Davis 
2:30 - 2:45pm EDTCoffee/Tea Break  
2:45 - 3:30pm EDTUsing Mixed Precision in Numerical Computations to Speedup Linear Algebra Solvers - Jack Dongarra, University of Tennessee - Oak Ridge National Lab 
3:45 - 4:30pm EDTPotential Directions for Moving IEEE-754 Forward - Jason Riedy, Georgia Institute of Technology 
4:45 - 5:00pm EDTClose the day  
Friday, May 8, 2020
TimeEventLocationMaterials
10:00 - 10:45am EDTThe Multiprecision Effort in the US Exascale Computing Project - Hartwig Anzt, University of Tennessee - Karlsruhe Institute of Technology 
11:00 - 11:15am EDTCoffee/Tea Break  
11:15 - 12:00pm EDTHardware and FPGAs computing just right - Florent de Dinechin, INSA Lyon 
12:15 - 1:30pm EDTBreak for Lunch / Free Time  
1:30 - 2:15pm EDTUsing ZFP Lossy Compression in HPC- Theoretical Support for the New Floating-Point Representation - Alyson Fox, Lawrence Livermore National Laboratory 
2:30 - 2:45pm EDTCoffee/Tea Break  
2:45 - 3:30pm EDTAutomatic Mixed Precision Analysis and Transformation Tool - Harshitha Menon, Lawrence Livermore National Laboratory 
3:45 - 4:30pm EDTThe Grass is Really Green on the Other Side - Empirical vs. Rigorous in Floating-Point Precision Analysis, Reproducibility, Resilience - Ganesh Gopalakrishnan, University of Utah 
4:45 - 5:00pm EDTClose the workshop  

Lecture Videos

The Grass is Really Green on the Other Side - Empirical vs. Rigorous in Floating-Point Precision Analysis, Reproducibility, Resilience

Ganesh Gopalakrishnan
University of Utah
May 8, 2020
VIRTUAL ONLY: Variable Precision in Mathematical and Scientific Computing
Play     Abstract     Slides

Automatic Mixed Precision Analysis and Transformation Tool

Harshitha Menon
Lawrence Livermore National Laboratory
May 8, 2020
VIRTUAL ONLY: Variable Precision in Mathematical and Scientific Computing
Play     Abstract

Using ZFP Lossy Compression in HPC- Theoretical Support for the New Floating-Point Representation

Alyson Fox
Lawrence Livermore National Laboratory
May 8, 2020
VIRTUAL ONLY: Variable Precision in Mathematical and Scientific Computing
Play     Abstract

Hardware and FPGAs computing just right

Florent de Dinechin
University of Lyon, INSA
May 8, 2020
VIRTUAL ONLY: Variable Precision in Mathematical and Scientific Computing
Play     Abstract     Slides

The Multiprecision Effort in the US Exascale Computing Project

Hartwig Anzt
Karlsruhe Institute of Technology
May 8, 2020
VIRTUAL ONLY: Variable Precision in Mathematical and Scientific Computing
Play     Abstract     Slides

Potential Directions for Moving IEEE-754 Forward

Jason Riedy
Georgia Institute of Technology
May 7, 2020
VIRTUAL ONLY: Variable Precision in Mathematical and Scientific Computing
Play     Slides

Using Mixed Precision in Numerical Computations to Speedup Linear Algebra Solvers

Jack Dongarra
University of Tennessee
May 7, 2020
VIRTUAL ONLY: Variable Precision in Mathematical and Scientific Computing
Play     Abstract     Slides

Dynamic Analyses for Floating-Point Precision Tuning

Cindy Rubio-González
University of California Davis
May 7, 2020
VIRTUAL ONLY: Variable Precision in Mathematical and Scientific Computing
Play

How Slow is Quadruple Precision

Paul Zimmermann
INRIA
May 7, 2020
VIRTUAL ONLY: Variable Precision in Mathematical and Scientific Computing
Play     Abstract     Slides

Variable precision computing- Applications and challenges

David Bailey
University of California, Davis
May 7, 2020
VIRTUAL ONLY: Variable Precision in Mathematical and Scientific Computing
Play     Abstract     Slides