Overcoming the challenges to feedback-directed optimization (Keynote Talk)

1 January 2000

conference paper
Published by Association for Computing Machinery (ACM)

Vol. 35 (7) , 1-11
https://doi.org/10.1145/351397.351408

Abstract

Feedback-directed optimization (FDO) is a general term used to describe any technique that alters a program's execution based on tendencies observed in its present or past runs. This paper reviews the current state of affairs in FDO and discusses the challenges inhibiting further acceptance of these techniques. It also argues that current trends in hardware and software technology have resulted in an execution environment where immutable executables and traditional static optimizations are no longer sufficient. It explains how we can improve the effectiveness of our optimizers by increasing our understanding of program behavior, and it provides examples of temporal behavior that we can (or could in the future) exploit during optimization.

Keywords

This publication has 23 references indexed in Scilit:

Static correlated branch prediction
ACM Transactions on Programming Languages and Systems, 1999
Procedure placement using temporal-ordering information
ACM Transactions on Programming Languages and Systems, 1999
Tempo
ACM Computing Surveys, 1998
The history of Fortran I, II, and III
IEEE Annals of the History of Computing, 1998
Changing interaction of compiler and architecture
Computer, 1997
Partial Evaluation Applied to Ray Tracing
Published by Springer Nature ,1996
The microarchitecture of superscalar processors
Proceedings of the IEEE, 1995
The superblock: An effective technique for VLIW and superscalar compilation
The Journal of Supercomputing, 1993
The multiflow trace scheduling compiler
The Journal of Supercomputing, 1993
The nonuniform distribution of instruction-level and machine parallelism and its effect on performance
IEEE Transactions on Computers, 1989