EDACC is a framework for the design, execution and analysis of large experiments with arbitrary algorithms on computer clusters. It automates many of the tasks commonly done manually such as distributing the algorithm executables across compute nodes, running the required computation tasks and collecting the results.
Core components
- Database: Stores information about algorithm executables, problem instances and experiments
- GUI Application: Java GUI application for managing algorithms and problem instances, creating and analysing experiments
- Computation Client: Execution of the algorithms on remote, distributed systems such as computer clusters
Optional components
- Web Frontend: Web client to perform monitoring and statistical analysis of experiment results
- Job server: Small application that increases the throughput of the system when large-scale grids are used
- Visualizer: Visualize the configuration process of algorithms and to analyze the results
- Automatic algorithm configurator API: Java API for parallel automatic algorithm configurators
- eAAC: Automatic configurator framework, which includes several automatic algorithm configurators
Releases
The latest release contains the GUI application, which can be used to set up an EDACC database, and the computation client. We recommend working with the latest development versions.
edacc-0.4.1.zip, 8.7M Sourcecode
Applications
EDACC was used to conduct the following solver competitions:
- SAT Challenge 2012 (Web Frontend)
- SAT Competition 2013 (Web Frontend)
- SAT Competition 2014 (Web Frontend)
- Synthesis Competition 2014
It was also used in tuning some of the solvers that participated in these competitions (see publications).
Publications
EDACC is developed at the Institute of Theoretical Computer Science of Ulm University and introduced in the following publications:
-
Adrian Balint, Daniel Gall, Gregor Kapler, Robert Retz, Daniel Diepold and Simon Gerber
EDACC - An advanced Platform for the Experiment Design, Administration and Analysis of Empirical Algorithms
to appear in the proceedings of LION5 -
Adrian Balint, Daniel Gall, Gregor Kapler, Robert Retz Experiment Design and Administration for Computer Cluster for SAT-solvers (EDACC), System Description
in JSAT Volume 7
Some references to EDACC in scientific work, publications and articles include:
-
Proceedings of SAT Challenge 2012; Solver and Benchmark Descriptions
-
Proceedings of SAT Competition 2013; Solver and Benchmark Descriptions
-
Proceedings of SAT Competition 2014; Solver and Benchmark Descriptions
-
Oliver Gableske, Sven Müelich, Daniel Diepold
On the Performance of CDCL-based Message Passing Inspired Decimation using Rho-Sigma-PMP-i
EasyChair Proceedings of the Pragmatics of SAT Workshop (POS13), 2013.
Dimetheus SAT Solver