Dietmar Ebner

Contacts

CD Laboratory Compilation Techniques for Embedded Processors Institute of Computer Languages E185/1 Vienna University of Technology Argentinierstraße 8 / 185 A-1040 Wien, Austria Fax: (+431) 58801-18598 Tel.: (+431) 58801-58521

Since August 2009, I am a postdoctoral researcher at the Lawrence Livermore National Laboratory (LLNL).

Dietmar Ebner is a Research Assistant at the Christian Doppler Laboratory Compilation Techniques for Embedded Processors. His research is focused on combinatorial optimizations for hard problems, in particular using dynamic programming, local search techniques, and integer linear programming. Current projects are on retargetable compiler backends and automatic parallelization for superscalar and VLIW architectures.

He received his M.Sc. (Dipl.Ing.) in 2005 from the Vienna University of Technology for his master thesis on exact crossing minimization for graphs using integer linear programming. Thereon, he spent six month at the Mitsubishi Electric Research Laboratories working on scheduling algorithms for hard transportation problems such as group elevator control. Since 2006, he is working on a project on optimizing compilers for embedded architectures that is partially funded by ON DEMAND Microelectronics.

Research Interests

• Compilation techniques and compiler optimizations
• Combinatorial optimization (local search techniques, ILP, DP, B&B)
• Efficient datastructures and algorithms
• Graph drawing and graph algorithms in general

VU2 185.324 Compilation Techniques for VLIW Architectures SS08, Florian Brandner and Dietmar Ebner, TUWIS++ Pages The course gives an overview of VLIW design principles and captures several real world examples along with their peculiarities compared to the general purpose computing domain. The main focus is on compilation techniques for ILP compilers, in particular speculation and predication, region scheduling techniques and software pipelining.

	Progressive Spill Code Placement Dietmar Ebner, Bernhard Scholz, Andreas Krall Accepted for CASES09 (International Conference on Compilers, Architecture, and Synthesis for Embedded Systems, Grenoble, France)	To Appear
	Generalized Instruction Selection using SSA-Graphs Dietmar Ebner, Florian Brandner, Bernhard Scholz, Andreas Krall, Peter Wiedermann, and Albrecht Kadlec In Proc. of LCTES08 (International Conference on Languages, Compilers, and Tools for Embedded Systems, Tucson, AZ)	[slides]
	Compiler Generation from Structural Architecture Descriptions Florian Brandner, Dietmar Ebner, and Andreas Krall In Proc. of CASES07 (International Conference on Compilers, Architecture, and Synthesis for Embedded Systems, Salzburg, Austria)
	Leveraging Predicated Execution for Multimedia Processing Dietmar Ebner, Florian Brandner, and Andreas Krall In Proc. of ESTIMEDIA07 (5th International Workshop on Embedded Systems for Real-Time Multimedia, Salzburg, Austria)
	A Branch-and-Cut Approach to the Crossing Number Problem C. Buchheim, M. Chimani, D. Ebner, C. Gutwenger, M. Jünger, G.W. Klau, P. Mutzel, and R. Weiskircher Discrete Optimization, Vol. 5(2), Special Issue in memory of George B. Dantzig, pp. 373-388, 2008
	Static Verification of Global Heap References in Java Native Libraries Andreas Krall, Christian Thalinger, Dietmar Ebner, and Florian Brandner In SPACE06 (Third Workshop on Semantics, Program Analysis, and Computing Environments for Memory Management, Charleston, South Carolina)
	Exact Crossing Minimization Christoph Buchheim, Dietmar Ebner, Michael Jünger, Gunnar W. Klau, Petra Mutzel, and René Weiskircher In Proc. of GD2005 (13th International Symposium on Graph Drawing, Limerick, Ireland)
	Handbook of Graph Drawing and Visualization, Roberto Tamassia (Editor) Chapter Crossings and Planarization Christoph Buchheim, Dietmar Ebner, Carsten Gutwenger, Michael Jünger, and Petra Mutzel	To Appear
	Optimal Crossing Minimization using Integer Linear Programming Dietmar Ebner Masters Thesis, Vienna University of Technology
	Label Number Maximization in the Slider Model Dietmar Ebner, Gunnar W. Klau, and René Weiskircher In Proc. of GD2004 (12th International Symposium on Graph Drawing, New York, USA)

	ACM SIGPLAN	Special Interest Group on Programming Languages SIGPLAN is a Special Interest Group of ACM that focuses on Programming Languages.
	ACM SIGBED	Special Interest Group on Embedded Systems SIGPLAN is a Special Interest Group of ACM that focuses on all aspects of Embedded Systems including both software and hardware.
	CGO 2009	International Symposium on Code Generation and Optimization (CGO) Seattle, Washington, March 22 - 25, 2009
	ETAPS 2009	European Joint Conferences on Theory and Practice of Software York, UK, March 22 - 29, 2009 CC International Conference on Compiler Construction
	PLDI 2009	ACM SIGPLAN Conference on Programming Language Design and Implementation Trinity College, Dublin, June 15-21, 2009

Curriculum Vitae

My résumé is available in PDF format!

Handy Stuff

• Cheat-Sheets for emacs, gdb, and the MIPS R2000 instruction set.
• A Collection of Computer Science Bibliographies
• LEO english dictionary
• Libraries
  • Institute of Computer Languages
  • Vienna University of Technology
  • University of Vienna
• CiteSeer Scientific Literature Digital Library
• Various online resources & documentation
  • The Not So Short Introduction to LaTeX
  • gcc
  • libstdc++
  • libxml++
  • A STL Programmers Guide
  • A C/C++ reference
  • Java 2 API

GCC tree dumps

GCC does not (yet) have a serializable intermediate representation. There is some work going on in the link time optimizations brach. Have also a look at the GCC Wiki and the Bugzilla database!

In the meantime, you can use this patch to create a GIMPLE dump after a particular phase in XML format by passing -fdump-tree-<phase>-xml. The patch should cleanly apply to gcc 4.1.

Force Directed Map Labeling

Sources, precompiled JAR archives, and benchmark instances for various algorithms for the map labeling problem in the (most general) four-slider model. All sources are available under the terms of the GNU GPL licence.