作者: M Alvanos , E Ayguade , RM Badia , V Beltran , A Bilas
DOI: 10.1109/MM.2010.94
关键词:
摘要: This article evaluates the scalability and productivity of six parallel programming models for heterogeneous architectures, finds that task-based using code data annotations require minimum effort while sustaining nearly best performance. However, achieving this result requires both extensions to control locality granularity proper interoperability with platform-specific optimizations.