Early Graph-based Resource Utilization Modeling
Bachelor’s Thesis / Student Research Project
Abstract
Abstract modeling of HW/SW systems is a relatively new research topic. This technique aims to capture only the essential parameters of software and hardware that influence their timing behavior. Data-flow graphs of software used in high-level synthesis explicitly model how much software can be parallelized.
This student project’s goal is to use data-flow graphs and hardware resource constraints to get an early runtime and resource utilization estimation for the execution of neural networks on machine learning accelerator hardware.
An Example of a data-flow graph and its resource utilization estimation:
Requirements
- Python
- Successfully atteded the lecture “Grundlagen der Rechnerarchitektur” and/or “Entwurf und Synthese Eingebetteter Systeme” (optional)
- Linux (optional)