Profile-Driven Automated Mixed Precision

06/01/2016
by   Ralph Nathan, et al.
0

We present a scheme to automatically set the precision of floating point variables in an application. We design a framework that profiles applications to measure undesirable numerical behavior at the floating point operation level. We use this framework to perform mixed precision analysis to heuristically set the precision of all variables in an application based on their numerical profiles. We experimentally evaluate the mixed precision analysis to show that it can generate a range of results with different accuracy and performance characteristics.

READ FULL TEXT
research
01/26/2016

Vectorization of Multibyte Floating Point Data Formats

We propose a scheme for reduced-precision representation of floating poi...
research
02/17/2021

NEAT: A Framework for Automated Exploration of Floating Point Approximations

Much recent research is devoted to exploring tradeoffs between computati...
research
11/10/2015

BOAT: a cross-platform software for data analysis and numerical computing with arbitrary-precision

BOAT is a free cross-platform software for statistical data analysis and...
research
10/25/2021

Mixed precision in Graphics Processing Unit

Modern graphics computing units (GPUs) are designed and optimized to per...
research
03/29/2022

Precision-aware Deterministic and Probabilistic Error Bounds for Floating Point Summation

We analyze the forward error in the floating point summation of real num...
research
10/28/2019

Adaptive Loss Scaling for Mixed Precision Training

Mixed precision training (MPT) is becoming a practical technique to impr...

Please sign up or login with your details

Forgot password? Click here to reset