Deploying the existing single-core application to multi and many-core platforms did not bring increase of the performance as it was expected in the first place where the multi-core overheads were not considered. The deployment of complex real-time systems with everyday increasing demands and possibilities is a challenging task for engineers when performance and efficiency have to be maximized and cost have to be minimized at the same time.
For already designed single core applications, it became necessary to provide automatic parallelization for multi and many core applications which will fulfill all of the previously defined requirements (e.g. fulfillment of deadlines as in single-core application). Description of the software application through model-based approach provides clearer overview of the inner software dependencies (e.g. communication between the runnables) and allows simulation of real-time scheduling.
Through representation of the application by the model-based approach different parallelization scenarios can be evaluated and adopted for complex real-time systems. Parallelization approach, presented here, is done based on the analysis of inner software dependencies and determination of the "parallelizable" software application parts. Whole process is automatized to perform evaluation of different parallel execution scenarios in order to determine the most performant parallel version of the initial application. Timing behavior of such applications is evaluated with event-based simulation tool. Simulation by tool of the parallel real-time application execution and their task's scheduling allows to evaluate different benefits and the trade-offs due necessary synchronization of the software parts, for example.
Skills
Timing, Model based Approach, Real-Time Systems, Multi and Many-Core Systems
Lernziele
Participants will get closer to the concepts of the model based approach and tooling in order to fasten up the development approaches.
Also, participants will be aware of the challenges and the unwanted effects that multi-core brings.
Use-cases will describe the challenges seen in the automotive area, but the principles can be mirrored also in other domains.
// Referent
//
Erna Oklapi
After finishing graduate studies at Electro-Technical Faculty, University of Belgrade, Serbia, Erna Oklapi studied Master of Applied Research in Engineering Sciences at University of Applied Sciences in Regensburg (OTH Regensburg). During the master studies, she was Research Associate in Laboratory for Safe and Secure Systems. She won a price from German Engineers Association (Verein Deutscher Ingenieure) for her master thesis for 2015. Currently she is working as a Technical Consultant at Timing-Architect Embedded Systems on one side and on the other side as Research Assistant in Laboratory for Safe and Secure Systems where she is working on her PhD. Her interests are multi and many-core real-time embedded systems with special focus on automatization and definition of parallelization.