Using machines to learn method-specific compilation strategies

Authors:
Ricardo Nabinger Sanchez;Jose Nelson Amaral;Duane Szafron;Marius Pirvu;Mark Stoodley
Affiliations:
University of Alberta, Edmonton, AB, Canada;University of Alberta, Edmonton, AB, Canada;University of Alberta, Edmonton, AB, Canada;IBM Toronto Software Laboratory, Markham, ON, Canada;IBM Toronto Software Laboratory, Markham, ON, Canada
Venue:
CGO '11 Proceedings of the 9th Annual IEEE/ACM International Symposium on Code Generation and Optimization
Year:
2011

Citing 20
Cited 2

Support-Vector Networks

Machine Learning
Fast, effective code generation in a just-in-time Java compiler

PLDI '98 Proceedings of the ACM SIGPLAN 1998 conference on Programming language design and implementation
Adaptive optimization in the Jalapeño JVM

OOPSLA '00 Proceedings of the 15th ACM SIGPLAN conference on Object-oriented programming, systems, languages, and applications
An Empirical Study of Selective Optimization

LCPC '00 Proceedings of the 13th International Workshop on Languages and Compilers for Parallel Computing-Revised Papers
Java(TM) Language Specification, The (3rd Edition) (Java (Addison-Wesley))

Java(TM) Language Specification, The (3rd Edition) (Java (Addison-Wesley))
Advanced Programming in the UNIX(R) Environment (2nd Edition)

Advanced Programming in the UNIX(R) Environment (2nd Edition)
Online performance auditing: using hot optimizations without getting burned

Proceedings of the 2006 ACM SIGPLAN conference on Programming language design and implementation
The DaCapo benchmarks: java benchmarking development and analysis

Proceedings of the 21st annual ACM SIGPLAN conference on Object-oriented programming systems, languages, and applications
Method-specific dynamic compilation using logistic regression

Proceedings of the 21st annual ACM SIGPLAN conference on Object-oriented programming systems, languages, and applications
Linux Kernel Development (2nd Edition) (Novell Press)

Linux Kernel Development (2nd Edition) (Novell Press)
Fast compiler optimisation evaluation using code-feature based performance prediction

Proceedings of the 4th international conference on Computing frontiers
Microarchitecture Sensitive Empirical Models for Compiler Optimizations

Proceedings of the International Symposium on Code Generation and Optimization
Rapidly Selecting Good Compiler Optimizations using Performance Counters

Proceedings of the International Symposium on Code Generation and Optimization
JavaTM just-in-time compiler and virtual machine improvements for server and middleware applications

VM'04 Proceedings of the 3rd conference on Virtual Machine Research And Technology Symposium - Volume 3
Cole: compiler optimization level exploration

Proceedings of the 6th annual IEEE/ACM international symposium on Code generation and optimization
A sequential dual method for large scale multi-class linear svms

Proceedings of the 14th ACM SIGKDD international conference on Knowledge discovery and data mining
LIBLINEAR: A Library for Large Linear Classification

The Journal of Machine Learning Research
Quick and Practical Run-Time Evaluation of Multiple Program Optimizations

Transactions on High-Performance Embedded Architectures and Compilers I
Automated just-in-time compiler tuning

Proceedings of the 8th annual IEEE/ACM international symposium on Code generation and optimization
Using Support Vector Machines to Learn How to Compile a Method

SBAC-PAD '10 Proceedings of the 2010 22nd International Symposium on Computer Architecture and High Performance Computing

Performance potential of optimization phase selection during dynamic JIT compilation

Proceedings of the 9th ACM SIGPLAN/SIGOPS international conference on Virtual execution environments
Exploring single and multilevel JIT compilation policy for modern machines 1

ACM Transactions on Architecture and Code Optimization (TACO)

Quantified Score

Hi-index	0.00

Visualization

Abstract

Support Vector Machines (SVMs) are used to discover method-specific compilation strategies in Testarossa, a commercial Just-in-Time (JiT) compiler employed in the IBM® J9 Java™ Virtual Machine. The learning process explores a large number of different compilation strategies to generate the data needed for training models. The trained machine-learned model is integrated with the compiler to predict a compilation plan that balances code quality and compilation effort on a per-method basis. The machine-learned plans outperform the original Testarossa for start-up performance, but not for throughput performance, for which Testarossa has been highly hand-tuned for many years.