期刊文献+

多寄存器组网络处理器上的寄存器分配技术 被引量:1

Register Allocation on Network Processors with Multiple Register Banks
下载PDF
导出
摘要 针对传统的图着色寄存器分配算法不能直接处理网络处理器的操作问题,提出了一种多寄存器组网络处理上的寄存器分配技术.在依次分析了一个符号寄存器可能位于哪些寄存器组?如果没有候选组,该如何解决这种冲突?如果有多个候选组,该选用哪个组等问题的基础上,通过将这些方法与图着色寄存器分配算法相融合,在IXP上实现了这种多寄存器组的寄存器分配,提高了它的可编程性.这种方法也可运用到其它具有类似寄存器结构的处理器上. To meet the requirements of network applications, some network processors, like Intel IXP series, apply an unconventional multiple-register-bank architecture. Register banks are connected by data paths such that data can be moved around in the register banks. Moreover, the source or destination operands of an instruction can reside in one or more register banks. Two sources selection rules further restrict the bank selection of the two source operands of an instruction. Facing these new problems, the classical graph coloring register allocation algorithm cannot be directly applied. A series of algorithms are presented to address the problems by answering the following questions. Which legal candidate register banks can a symbolic register reside in? If the candidate banks are empty, how to resolve the bank conflicts? And if there're multiple candidate banks, which bank should be selected? After integrating these algorithms with the graph-coloring register allocation, authors have implemented a multiple-register-bank register allocator on IXP, which improved the programmability of IXP. The algorithms can also be applied to other microprocessors with the similar register architecture.
出处 《计算机学报》 EI CSCD 北大核心 2006年第1期66-72,共7页 Chinese Journal of Computers
关键词 寄存器分配 网络处理器 多寄存器组 register allocation network processor, multiple register banks
  • 相关文献

参考文献14

  • 1http://www.radisys.com.
  • 2http://www.ampl.com.
  • 3http://www.cplex.com.
  • 4Crowley P, , Franklin M. A. et al., Network Processor Design Issues and Practices. Volume 1. CA, USA: Morgan Kaufmann Publisher, 2003.
  • 5Intel IXP2800 Nelwork Processor Hardware Reference Manual, Nov. 2002.
  • 6Intel IXP2400/IXP2800 Network Processor Programmer's Reference Manual, Jan. 2003.
  • 7Chaitin G. J.. Register allocation and spilling via graph coloring. In: Proceedings of the SIGPLAN'82 Symposium on Compiler Construction, Boston, MA, USA, 1982, 98-101.
  • 8Bernstein D. , Cohen D. et al.. Code duplication: An assist for global instruction scheduling. In: Proceedings of Micro24,IEEE Computer Society, Albuquerque, New Mexico, Puerto Rico, 1991, 103-113.
  • 9Zhuang X. T. , Pande S.. Resolving register bank conflicts for a network processor. In: Proceedings of the International Conferenee on Parallel Architecture and Compilation Techniques,New Orleans, LA, USA, 2003, 269-278.
  • 10Chandra P. R. , Hady F. et al. , Benchmarking network processors. In: Proceedings of the 1st Workshop on Network Processors, Cambridge, MA, 2002, 426-433.

同被引文献5

引证文献1

二级引证文献1

相关作者

内容加载中请稍等...

相关机构

内容加载中请稍等...

相关主题

内容加载中请稍等...

浏览历史

内容加载中请稍等...
;
使用帮助 返回顶部