Sprungmarken

Servicenavigation

Hauptnavigation

Sie sind hier:

Hauptinhalt

Detailseite



Hardware/Software Simulation Environment for Automotive Multicore Systems

Chair: Embedded Systems

Supervisior: Prof. Dr.-Ing. Selma Saidi,

Start: Sommersemester 2020 - 01.04.2020

Maximum amount of participants: 6

Description: Embedded hardware platforms currently used in automotive are becoming increasingly complex and heterogeneous with the adoption of multi/many-core processors systems. Appropriate simulation frameworks allow to understand the timing behavior of the system depending on software and hardware characteristics, as well as explore different design architectural choices and quantitatively evaluate their implications on system performance.

The goal of the project is to build a system simulation framework in Gem5 of an SoC platform used in automotive. The specification of the platform as well as the application are provided in AMALTHEA, an open source model-based design and analysis framework used by industry to exchange high-level automotive software and hardware models.








Tasks:
- Create a framework of an automotive system-on-chip hardware using Gem5 including multiple embedded processors, local interconnects and memory hierarchies, ​
- Synthesize the Gem5 components from an AMALTHEA SW and HW description​,
- Create variants of the considered models with variable configurations such as different applications loads, different processors architectures and different memory components including caches or SPM, DMAs and DRAMs. ​
- Evaluate the performance of different configurations.​

You will learn:
- Some characteristics of automotive applications and automotive hardware,​
- How to build and simulate your own computer architecture based on existing architectural building blocks, ​
- How to investigate different architectural design alternatives, ​
- The implications of hardware architectural features on system performance.

Requirements:
- Basic notions of computer architectures,
- Programming experience in python/C++, Java.


Note: this project work will be done in collaboration with Bosch Corporate Research (with Falk Rehm and Dr. Arne Hamann).

Contact:
selma.saidi@tu-dortmund.de
Otto-Hahn-Str. 4, Campus Nord
Roum p1-04-217


 

The project group takes place!

Participants: 215657, 0, 215176, 223722, 215191, 215899,