SPREX: A Soft Processor with Runahead Execution

被引:0
作者
Aasaraai, Kaveh [1 ]
Moshovos, Andreas [1 ]
机构
[1] Univ Toronto, Dept Elect & Comp Engn, Toronto, ON, Canada
来源
2012 INTERNATIONAL CONFERENCE ON RECONFIGURABLE COMPUTING AND FPGAS (RECONFIG) | 2012年
关键词
soft processor; Runahead; high performance;
D O I
暂无
中图分类号
TP3 [计算技术、计算机技术];
学科分类号
0812 ;
摘要
There is a growing demand for high-performance computation cores in embedded devices built over reconfigurable hardware. As a result, various soft core architecture techniques have been proposed, each targeting different application classes. This work presents SPREX, an FPGA-friendly Runahead soft processor architecture that targets applications with unstructured instruction level parallelism. The architecture of choice for such applications has traditionally relied on a mix of superscalar, out-of-order, and speculative execution. Unfortunately, the implementation of these techniques does not map well on reconfigurable hardware. This work shows that by exploiting the key characteristics of reconfigurable fabrics, and by tuning the architecture for the embedded environment, a fast and practical Runahead soft processor is viable. Runahead has been shown to offer many of the benefits of conventional architectures for the applications this work targets. We show that the proposed Runahead architecture improves performance of a simple 5-stage pipeline by 9% on the average and by as much as 36%.
引用
收藏
页数:7
相关论文
共 15 条
[1]  
A. Corp, 2010, STRAT 3 DEV HDB
[2]  
Aasaraai K, 2010, P INT C RECONFIGURAB
[3]  
Aasaraai K., 2009, 19 INT C FIELD PROGR
[4]   NCOR: An FPGA-Friendly Nonblocking Data Cache for Soft Processors with Runahead Execution [J].
Aasaraai, Kaveh ;
Moshovos, Andreas .
INTERNATIONAL JOURNAL OF RECONFIGURABLE COMPUTING, 2012, 2012
[5]  
Altera Corp, 2010, NIOS 2 PROC REF HDB
[6]  
[Anonymous], ALT DE3 DEV SYST STR
[7]  
Dundas J., 1997, Conference Proceedings of the 1997 International Conference on Supercompting, P68, DOI 10.1145/263580.263597
[8]  
Fytraki S., 2008, Design and Automation Europe
[9]   Self-optimizing memory controllers:: A reinforcement learning approach [J].
Ipek, Engin ;
Mutlu, Onur ;
Martinez, Jose F. ;
Caruana, Rich .
ISCA 2008 PROCEEDINGS: 35TH INTERNATIONAL SYMPOSIUM ON COMPUTER ARCHITECTURE, 2008, :39-+
[10]  
Mesa-Martinez F.J., 2006, Workshop on Architectural Research Prototyping (WARP), held in conjunction with ISCA-33, P61