Research: Vertical Profiling

Vertical Profiling is an approach to performance understanding in modern computer systems. It enables the correlation of performance and behavior information accross the layers of modern systems (hardware, operating system, virtual machine, application server, and application) to identify causes of performance problems.

Vertical Profiling is the topic of my PhD thesis. My advisor is Amer Diwan. We collaborate with the PEM (Performance and Environment Monitoring) group in the PERCS (Productive, Easy-to-use, Reliable Computing System) project at IBM Research. And we are part of the DaCapo (Dynamic Cooperative Performance Optimization for Java) project.

Dissertation

• Understanding Program Performance Using Temporal Vertical Profiling
  Matthias Hauswirth
  Doctoral Dissertation, 2005
  [ PDF ]

Conference Publications

• Vertical Profiling: Understanding the Behavior of Object-Oriented Applications
  Matthias Hauswirth, Peter F. Sweeney, Amer Diwan, Michael Hind
  19th Annual ACM SIGPLAN Conference on Object-Oriented Programming, Systems, Languages, and Applications - OOPSLA 2004
  [ PDF ]
  Copyright ACM, 2004. This is the author's version of the work. It is posted here by permission of ACM for your personal use. Not for redistribution. The definitive version will appear in the ACM SIGPLAN 18th Conference on Object Oriented Programming Systems Languages and Applications (OOPSLA'04), Oct. 24-28, 2004.
• Low-Overhead Memory Leak Detection Using Adaptive Statistical Profiling
  Trishul Chilimbi, Matthias Hauswirth
  11th International Conference on Architectural Support for Programming Languages and Operating Systems - ASPLOS 2004
  [ PDF | PS ]
  Copyright ACM, 2004. This is the author's version of the work. It is posted here by permission of ACM for your personal use. Not for redistribution. The definitive version will appear in the ACM 11th International Conference on Architectural Support for Programming Languages and Operating Systems (ASPLOS'04), Oct. 9-13, 2004.
• Performance and Environment Monitoring for Whole-System Characterization and Optimization
  Robert W. Wisniewski, Peter F. Sweeney, Kartik Sudeep, Matthias Hauswirth, Evelyn Duesterwald, Calin Cascaval, Reza Azimi
  1st Watson Conference on Interaction between Architecture, Circuits, and Compilers - P=ac² 2004
  [ PDF ]
• Using Hardware Performance Monitors to Understand the Behavior of Java Applications
  Peter F. Sweeney, Matthias Hauswirth, Brendon Cahoon, Perry Cheng, Amer Diwan, David Grove, Michael Hind
  3rd Virtual Machine Research and Technology Symposium - VM 2004
  [ PDF ]
  First published in Proceedings of VM'04.
• Static Load Classification for Improving the Value Predictability of Data Cache Misses
  Martin Burtscher, Amer Diwan, Matthias Hauswirth
  Conference on Programming Language Design and Implementation - PLDI 2002
  [ PDF | PS ]
  Copyright by the Association for Computing Machinery, Inc. Permission to make digital or hard copies of part or all of this work for personal or classroom use is granted without fee provided that copies are not made or distributed for profit or commercial advantage and that copies bear this notice and the full citation on the first page. Copyrights for components of this work owned by others than ACM must be honored. Abstracting with credit is permitted. To copy otherwise, to republish, to post on servers, or to redistribute to lists, requires prior specific permission and/or a fee. Request permissions from Publications Dept, ACM Inc., fax +1 (212) 869-0481, or permissions@acm.org.

Workshops & Demonstrations

• A Case for Vertical Profiling
  Matthias Hauswirth, Peter Sweeney, Amer Diwan, Michael Hind
  2nd Workshop on Managed Runtime Environments - MRE 2004
  
• The Need for a Whole-System View of Performance
  Matthias Hauswirth, Peter Sweeney, Amer Diwan, Michael Hind
  2nd Workshop on Component and Middleware Performance - CMP 2004
  
• Performance Explorer: Understanding Java Application Behavior
  Matthias Hauswirth, Peter Sweeney, Amer Diwan
  OOPSLA Demonstration - OOPSLA 2004
  

Tools

To enable my research into vertical profiling I have built a visualization and analysis tool called Performance Explorer (PE). PE is currently being used for the analysis of traces generated by the Jikes RVM virtual machine, and by the K42 open source operating system.

PE is going to be released as part of the Jikes RVM open source project soon. See Jikes RVM CVS repository.

Links

Some links relevant to my research activities.