Custom-Precision Mathematical Library Explorations for Code Profiling and Optimization
| Autor: | Matei Istoan, Eric Petit, David Defour, Pablo de Oliveira Castro |
|---|---|
| Přispěvatelé: | LAboratoire de Mathématiques et PhySique (LAMPS), Université de Perpignan Via Domitia (UPVD), Parallélisme, Réseaux, Systèmes, Modélisation (PRISM), Université de Versailles Saint-Quentin-en-Yvelines (UVSQ)-Centre National de la Recherche Scientifique (CNRS), Laboratoire d'Informatique Parallélisme Réseaux Algorithmes Distribués (LI-PaRAD), Université de Versailles Saint-Quentin-en-Yvelines (UVSQ), Intel Corporation [USA], Universite de Versailles, Intel corporation, Universite de Perpignan, Exascale Computing Research |
| Rok vydání: | 2020 |
| Předmět: |
Optimization
FOS: Computer and information sciences Mathematical library Computer science [INFO.INFO-GL]Computer Science [cs]/General Literature [cs.GL] Binary number 010103 numerical & computational mathematics 02 engineering and technology Satellite tracking 01 natural sciences Index Terms-HPC Floating-Point Arithmetic mathematical library libm 0202 electrical engineering electronic engineering information engineering Elementary function 0101 mathematics Implementation Profiling (computer programming) custom-precision [INFO.INFO-AO]Computer Science [cs]/Computer Arithmetic Energy consumption 020202 computer hardware & architecture Computer engineering Computer Science - Mathematical Software Mathematical Software (cs.MS) Specialization |
| Zdroj: | ARITH [Research Report] Universite de Versailles; Intel corporation; Universite de Perpignan; Exascale Computing Research. 2020 |
| DOI: | 10.1109/arith48897.2020.00026 |
| Popis: | The typical processors used for scientific computing have fixed-width data-paths. This implies that mathematical libraries were specifically developed to target each of these fixed precisions (binary16, binary 32, binary64). However, to address the increasing energy consumption and throughput requirements of scientific applications, library and hardware designers are moving beyond this one-size-fits-all approach. In this article we propose to study the effects and benefits of using user-defined floating-point formats and target accuracies in calculations involving mathematical functions. Our tool collects input-data profiles and iteratively explores lower precisions for each call-site of a mathematical function in user applications. This profiling data will be a valuable asset for specializing and fine-tuning mathematical function implementations for a given application. We demonstrate the tool’s capabilities on SGP4, a satellite tracking application. The profile data shows the potential for specialization and provides insight into answering where it is useful to provide variable-precision designs for elementary function evaluation. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|