PDF-Version

Description

In today's technology nodes a growing number of chips contain safety-critical components which are used in advanced driver assistance and in future for level 5 full self driving vehicles. The chips used in this area have demanding requirements for long-term reliability which increases the simulation runtime for test validation. Therefore eletronic design automation tools are mostly compute intensive when running circuit simulations and result therefore in a bottleneck, when simulated with multi-million gate designs.

Problem statement

This bachelor thesis focus on implementing an optimized greedy algorithm in C++ that groups parallel computable faults together and therefore optimizes both the runtime of the greedy algorithm and the runtime of the GPU-accelerated fault simulation. Figure 1 shows a simple example of this idea. The evaluation of this work is done by comparing the new implemented algorithm with an existing approach implemented in Java [2].

Solution aspects

• Literature survey on the problem
• Implement and evaluate both approaches
• Using GPU-Simulation
• Evaluation of the approach by means of simulation

Requirements

• Interest in working in a current research project FAST supported by the DFG
• Skills in C++, git, cmake, (a little bit Java)

Literature

• E. Schneider, "Mulit-level simulation of nano-electronic digital circuits on GPUs" - Dissertation, University of Stuttgart 2019.
• E. Schneider, S. Holst, M. A. Kochte, X. Wen and H. Wunderlich, "GPU-accelerated small delay fault simulation", 2015 Design, Automation & Test in Europe Conference & Exhibition (DATE), Grenoble, France, 2015, pp. 1174-1179.

Contact

Jan Dennis Reimer, M.Sc.
E-Mail: jan.dennis.reimer@uni-paderborn.de
Telefon: 05251 60-3922
Raum: P1.6.08.4