FPBench Logo

FPTalks 2021

The leading edge of floating-point research

The floating-point research community.
Online. July 14, 2021.

FPTalks 2021 was held online on Zoom. All talks were live-streamed and recorded on Youtube, accessible via the links below. Talks were be 10 minutes long, followed by audience questions over Slack.

Session 1 (Youtube)


Pavel Panchekha, University of Utah

A Two-Phase Approach for Conditional Floating-Point Verification

Debasmita Lohar, Max Planck Institute

SecFloat: Bridging Floating-Point and Secure 2-Party Computation

Deevashwer Rathee, Microsoft Research India

Synthesizing Efficient Low-Precision Kernels

Anastasiia Izycheva, Technical University of Munich


Session 2 (Youtube)

Python v. IEEE: Special cases aren't special enough to break the rules

Mark Dickinson, Enthought

Generating guard-stable C code with PRECiSA

Laura Titolo, National Aeronautics and Space Administration

Deductive Verification of Floating-Point Java Programs in KeY

Rosa Abbasi, Max Planck Institute

Combining Precision Tuning and Rewriting

Brett Saiki, University of Washington


Session 3 (Youtube)

“Is that a right answer?” Evaluating correctness in million-line climate models

Allison Baker, National Center for Atmospheric Research

Fast and Efficient Bit-Level Precision Tuning

Dorra Ben Khalifa, Université Perpignan via Domitia

Implementing, optimizing, verifying and validating mathematical hardware

Theo Drane, Intel

Interval Constraint-Based Mutation Testing of Numerical Specifications

Clothilde Jeangoudoux, Max Planck Institute


Session 4 (Youtube)

Accuracy for Rendering: as much as possible, please (so long as it's free)

Matt Pharr, NVIDIA

High Performance Correctly Rounded 32-bit Math Libraries

Jay P. Lim, Rutgers University

Compiler-Induced Numerical Variability in Scientific Applications

Ignacio Laguna, Lawrence Livermore National Laboratory

Rigorous Roundoff Error Analysis of Probabilistic Computations

Rocco Salvia, University of Utah


Session 5 (Youtube)

ZFP: Compressed Floating-Point Arrays

Peter Lindstrom, Lawrence Livermore National Laboratory

Seesaw: Robustness Analysis via Symbolic Automatic Differentiation

Tanmay Tirpankar, University of Utah

Application Driven Custom Number Systems

Theo Omtzigt, Stillwater Supercomputing

Parallel Shadow Execution to Accelerate the Debugging of Numerical Errors

Sangeeta Chowdhary, Rutgers University

Conclusion & Break

FPTalks 2020 was also held online and its schedule is archived.