Mon 15 Jun 2015 09:15 - 09:40 at PLDI Main BLUE (Portland 254-255) - Distinguished Papers Chair(s): Steve Blackburn

Scientific and engineering applications depend on floating point arithmetic to approximate real arithmetic. Unfortunately, this approximation introduces rounding error, which can accumulate to produce unacceptable results. While the numerical methods literature provides techniques to mitigate rounding error, applying these techniques requires manually rearranging expressions and understanding the finer details of floating point arithmetic.

We introduce Herbie, a tool which automatically improves floating point accuracy by searching for error-reducing transformations. Herbie estimates and localizes rounding error, applies a database of rules to generate improvements, takes series expansions, and combines improvements for different input regions. We evaluated Herbie on every example from a chapter in a classic numerical methods textbook, and found that Herbie was able to improve accuracy on each example, some by up to 60 bits, while imposing an average performance overhead of 11%. Colleagues in machine learning have applied Herbie to significantly improve the results of a clustering algorithm, and a mathematical library has accepted a patch generated using Herbie.

PLDI 2015 Artifact Evaluated Badge

Conference Day
Mon 15 Jun

Displayed time zone: Tijuana, Baja California change

09:00 - 11:00
Distinguished PapersResearch Papers at PLDI Main BLUE (Portland 254-255)
Chair(s): Steve BlackburnAustralian National University
09:00
15m
Day opening
Opening and Welcome
Research Papers
Steve BlackburnAustralian National University , David GroveIBM Research
09:15
25m
Talk
Automatically Improving Accuracy for Floating Point Expressions
Research Papers
Pavel PanchekhaUniversity of Washington, Alex Sanchez-SternUniversity of Washington, James R. WilcoxUniversity of Washington, Zachary TatlockUniversity of Washington, Seattle
Media Attached
09:40
25m
Talk
Diagnosing Type Errors with Class
Research Papers
Danfeng ZhangCornell University, Andrew C. Myers, Dimitrios VytiniotisMicrosoft Research, Cambridge, Simon Peyton JonesMicrosoft Research, Cambridge
Media Attached
10:05
25m
Talk
Provably Correct Peephole Optimizations with Alive
Research Papers
Nuno P. LopesMicrosoft Research, David MenendezRutgers University, Santosh NagarakatteRutgers University, John RegehrUniversity of Utah
Pre-print Media Attached
10:30
20m
Talk
One Minute Madness
Research Papers