Lecture Series of the Research Center HPC

October 23, 2019, 11:15
3W03, ICT Building 2nd floor,
Technikerstrasse 21a, 6020 Innsbruck

 

Insights in Automated Design Space Exploration for Embedded Systems
by Fedor Smirnov, Friedrich-Alexander Universität Erlangen-Nürnberg

Over the past decades, embedded processors have found their way into an ever increasing number of products. Advances in process technology downsizing enable the integration of an extensive number of processing elements on a single chip, so that embedded systems provide more processing power than ever before.

While this newly available processing power enables the implementation of novel applications with high computational requirements, the high number of processors and the complexity of the applications make a design by hand error-prone and expensive, if at all possible. Consequently, approaches for an automated design which are able to efficiently search through the space of feasible problem solutions and yield solutions with a high quality w.r.t. multiple---oftentimes non-linear and conflicting---design objectives are of utmost importance.

This talk will discuss possible ways to address current challenges in the area of design automation. In particular, it will demonstrate (a) how the design process can be modeled as a constrained combinatorial problem (CCP), (b) introduce SAT-Decoding, an optimization approach tailored to CCPs, and (c) discuss current research topics in this area.

Fedor Smirnov is a postdoctoral researcher at the System-level Design Automation group at Hardware/Software Co-Design, Friedrich-Alexander-Universität Erlangen-Nürnberg (FAU). He received both his bachelor’s and his master's degree in Mechatronics from the FAU, Germany, where he also defended his dissertation a few weeks ago.

His main research interests include multi-objective optimization of constrained problems as well as the design automation of embedded systems.

Nach oben scrollen