摘要: 指令级并行在提高处理器运行速度方面显得越来越重要,if转换技术是一种在处理器支持条件执行的前提下,有助于提高指令级并行度的编译优化技术。在详细分析GCC(GNU Compiler Collection)内部if转换技术的实现机制和算法的基础上,针对matrix体系结构特点,对GCC中现有if转换算法进行了移植与改进,实现了matrix编译器的if转换算法。实验证明,改进后的if转换算法能够更有效地移除分支,减少基本块的数量,扩大单个基本块的范围,有助于编译器生成更加优化的代码。
[1] 张晨曦,王志英,张春元,等.计算机体系结构(第二版)[M]. 北京:高等教育出版社,2005:79-80 [2] 田祖伟,赵克佳,汪小飞.GCC基于IA-64谓词执行的if转换技术研究[J].微电子学与计算机,2005,22(6):188-196 [3] Strtling A.Optimizing the GCC Suite for a VLIW Architecture.http://www.qucosa.de,2013 [4] Stallman R M.GNU Compiler Collection (GCC) Internals.http://gcc.gnu.org,2013 [5] Kumar R,Saxena A K,Singh P K.A Novel Heuristic for Selection of HyperBlock in If-Conversion[J].Electronics Computer Technology,2001,6:232-235 [6] Jacome M F,de Veciana G,Pillai S.Clustered VLIW Architectures with Predicated Switching[A]∥DAC ’01Proceedings of the 38th annual Design Automation Conference,2001[C].New York:Association for Computing Machinery,2001:696-701 [7] Park J C H,Schlansker M.On Predicated Execution.Software and Systems Laboratory.https://www.hpl.hp.com,2013 [8] Zimmerman E,Zilles C.On the Energy Effectiveness of If-conversion in Superscalar Microprocessors.http://www.cs.illinois.edu,2013 [9] Chuang Wei-haw,Calder B,Ferrante J.Phi-Predication for Li-ght-Weight if-Conversion[A]∥Proceedings of the International Symposium on Code Generation and Optimization:Code Generation and Optimization,CGO 2003.International Symposium on,2003[C].California,2003:179-190 [10] Monica S L,Robert P W.Limits of control flow on parallelism[A]∥ISCA ’92Proceedings of the 19th annual international symposium on Computer architecture,1992[C].New York:Association for Computing Machinery,1992:46-57 [11] Mahlke S A,Lin D C,Chen W Y,et al.Effective compiler support for predicated execution using the HyperBlock[A]∥MICRO 25Proceedings of the 25th Annual International Sympo-sium on Microarchitecture,1992[C].New York:Association for Computing Machinery,1992:45-54 |
No related articles found! |
|