Precision and Performance: Floating Point and Ieee 754 Compliance for Nvidia Gpus

Precision and Performance: Floating Point and Ieee 754 Compliance for Nvidia Gpus

Precision & Performance: Floating Point and IEEE 754 Compliance for NVIDIA GPUs

A number of issues related to floating point accuracy and compliance are a frequent source of confusion on both CPUs and GPUs. The purpose of this white paper is to discuss the most common issues related to NVIDIA GPUs and to supplement the documentation in the CUDA C Programming Guide [7].

Half-precision Floating-point Ray Traversal

Fraction (10 bits) Sign (1 bit) Exponent (5 bits) 16-bit floating-point format defined in IEEE 754-2008 standard Storage support on most of the modern CPUs and GPUs Native computation support especially on mobile platforms (Up coming nVidia Pascal desktop GPUs are announced to have native computation support) Pros: Smaller cache footprint (compared to "regular" 32-bit floats) More energy effici...

A Detailed Study of the Numerical Accuracy of GPU-Implemented Math Functions

Modern programmable GPUs have demonstrated their ability to significantly accelerate important classes of non-graphics applications; however, GPUs' substandard support for floating-point arithmetic can severely limit their usefulness for general-purpose computing. Current GPUs do not support double-precision computation and their single-precision support glosses over important aspects of the IE...

Efficient Implementation of Strict Floating-point Semantics

IA-32 processors yield different results of floating-point operations from other processors, even though they are all compliant with IEEE 754. The Java specifications need runtime systems to implement the FP-strict semantics, which other IEEE 754 compliant processors naturally fulfill. We implemented the semantics on a Java Just-in-Time compiler for IA-32. This study reveals that single-precisi...