A framework for reducing the cost of instrumented code
Authors:
Open Access
- 1 May 2001
- conference paper
- Published by Association for Computing Machinery (ACM)
- Vol. 36 (5) , 168-179
Abstract
Instrumenting code to collect proling information can causesubstantial execution overhead. This overhead makes instrumentationdicult to perform at runtime, often preventingmany known oine feedback-directed optimizations frombeing used in online systems. This paper presents a generalframework for performing instrumentation sampling toreduce the overhead of previously expensive instrumentation.The framework is simple and eective, using codeduplicationand counter-based sampling to allow ...Keywords
This publication has 25 references indexed in Scilit:
- Adaptive optimization in the Jalapeño JVM2000
- Practicing JUDO2000
- Dynamo2000
- The Jalapeño dynamic optimizing compiler for Java1999
- Fast, effective code generation in a just-in-time Java compiler1998
- Exploiting hardware performance counters with flow and context sensitive profiling1997
- Reconciling responsiveness with performance in pure object-oriented languagesACM Transactions on Programming Languages and Systems, 1996
- Optimally profiling and tracing programsACM Transactions on Programming Languages and Systems, 1994
- Profile‐guided automatic inline expansion for C programsSoftware: Practice and Experience, 1992
- Debugging optimized code with dynamic deoptimization1992