Study on the Mapping of Streaming Application on Many-Core Architecture

Streaming program is an important application of large-scale parallel processing. The common parallel method for streaming program is not suitable for many-core architecture. For the streaming programs, this paper proposes a novel mapping method based on hardwire architecture. The method uses SPM (Scratch-Pad Memory) to transfer the data. Thus, the number of accessing shared Cache is reduced and the cost of communication is also reduced. The experimental results show the effectiveness of the method.

[1]  David A. Wood,et al.  Managing Wire Delay in Large Chip-Multiprocessor Caches , 2004, 37th International Symposium on Microarchitecture (MICRO-37'04).

[2]  Dongrui Fan,et al.  Architectural Support and Evaluation of Cilk Language on Many-Core Architectures: Architectural Support and Evaluation of Cilk Language on Many-Core Architectures , 2009 .

[3]  Yun Zhang,et al.  Revisiting the Sequential Programming Model for Multi-Core , 2007, 40th Annual IEEE/ACM International Symposium on Microarchitecture (MICRO 2007).

[4]  William Thies,et al.  A Practical Approach to Exploiting Coarse-Grained Pipeline Parallelism in C Programs , 2007, 40th Annual IEEE/ACM International Symposium on Microarchitecture (MICRO 2007).

[5]  Ioannis E. Venetis,et al.  Optimizing the LU Benchmark for the Cyclops-64 Architecture , 2009 .

[6]  Michael I. Gordon,et al.  Exploiting coarse-grained task, data, and pipeline parallelism in stream programs , 2006, ASPLOS XII.

[7]  Long Guo.  Architectural Support and Evaluation of Cilk Language on Many-Core Architectures , 2008 .

[8]  Samuel Williams,et al.  The Landscape of Parallel Computing Research: A View from Berkeley , 2006 .

[9]  Lin Wei.  Using Write Mask to Support Hybrid Write-Back and Write-Through Cache Policy on Many-Core Architectures , 2008 .

[10]  Anoop Gupta,et al.  The SPLASH-2 programs: characterization and methodological considerations , 1995, ISCA.

[11]  Hao Zhang,et al.  Using Write Mask to Support Hybrid Write-Back and Write-Through Cache Policy on Many-Core Architectures: Using Write Mask to Support Hybrid Write-Back and Write-Through Cache Policy on Many-Core Architectures , 2009 .