Selective Register Renaming: A Compiler-Driven Approach to Dynamic Register Renaming

Authors:
N. Zingirian;Massimo Maresca
Affiliations:
-;-
Venue:
HPCN Europe 2001 Proceedings of the 9th International Conference on High-Performance Computing and Networking
Year:
2001

Citing 6
Cited 0

Advanced compiler optimizations for supercomputers

Communications of the ACM - Special issue on parallelism
Instruction Issue Logic for High-Performance, Interruptible, Multiple Functional Unit, Pipelined Computers

IEEE Transactions on Computers
Computer architecture (2nd ed.): a quantitative approach

Computer architecture (2nd ed.): a quantitative approach
Delaying physical register allocation through virtual-physical registers

Proceedings of the 32nd annual ACM/IEEE international symposium on Microarchitecture
The MIPS R10000 Superscalar Microprocessor

IEEE Micro
The Design Space of Register Renaming Techniques

IEEE Micro

Quantified Score

Hi-index	0.00

Visualization

Abstract

Dynamic register renaming is a mechanism present in many high performance microprocessors of latest generation aimed at removing false dependencies from the code. Unfortunately, in many cases, this mechanism keeps busy more registers than the necessary. In this paper we introduce a novel technique, called Selective Register Renaming, in which the compiler helps the processor to save physical registers when the hardware renames registers. The paper explains the principles of this technique and shows its effects on several Livermore Kernel Loops.