Code generators for mathematical functions - Sorbonne Université Access content directly
Preprints, Working Papers, ... Year : 2014

Code generators for mathematical functions

Olga Kupriianova
Nicolas Brunie
  • Function : Author
  • PersonId : 765150
  • IdRef : 178343544
Christoph Lauter

Abstract

A typical floating-point environment includes sup-port for a small set of about 30 mathematical functions such as exponential, logarithms and trigonometric functions. These functions are provided by mathematical software libraries (libm), typically in IEEE754 single, double and quad precision. This article suggests to replace this libm paradigm by a more general approach: the on-demand generation of numerical func-tion code, on arbitrary domains and with arbitrary accuracies. First, such code generation opens up the libm function space available to programmers. It may capture a much wider set of functions, and may capture even standard functions on non-standard domains and accuracy/performance points. Second, writing libm code requires fine-tuned instruction selec-tion and scheduling for performance, and sophisticated floating-point techniques for accuracy. Automating this task through code generation improves confidence in the code while enabling better design space exploration, and therefore better time to market, even for the libm functions. This article discusses, with examples, the new challenges of this paradigm shift, and presents the current state of open-source function code generators.
Fichier principal
Vignette du fichier
metalibm.pdf (1.51 Mo) Télécharger le fichier
Origin Files produced by the author(s)

Dates and versions

hal-01084726 , version 1 (19-11-2014)
hal-01084726 , version 2 (20-11-2014)

Identifiers

  • HAL Id : hal-01084726 , version 1

Cite

Olga Kupriianova, Nicolas Brunie, Florent de Dinechin, Christoph Lauter. Code generators for mathematical functions. 2014. ⟨hal-01084726v1⟩
575 View
2231 Download

Share

Gmail Mastodon Facebook X LinkedIn More