Modern Field Programmable Gate Arrays (FPGAs) offer a solution to several issues related to real-time on-board systems, such as guaranteed execution time. They are currently considered as target platforms for space applications. How-ever, the complexity of producing circuits on these components poses a challenge to their widespread adoption. To address this issue, high-level synthesis tools provide another layer of abstraction above the logic circuit design process, for example compiling C code into Hardware Description Languages such as VHDL or Verilog. However, high-level synthesis results are poorly predictable and do not guarantee the efficient use of recent FPGA capabilities provided by new primitives like digital signal processor or random-access memory. In this paper we propose a compilation chain dedicated to reactive systems, ie. controllers, providing a more predictable synthesis process for critical embedded control applications. The implemented solution demonstrates timing performance equivalent to the traditional synthesis process with a more predictable result.
A Reactive System-Specific Compilation Chain from Synchronous Dataflow Models to FPGA Netlist
15.07.2024
590689 byte
Aufsatz (Konferenz)
Elektronische Ressource
Englisch
Scheduled Dataflow Architecture: A Synchronous Execution Paradigm for Dataflow
British Library Online Contents | 1999
|Tabu Search: Implementation & Complexity Analysis for Netlist Partitioning
British Library Online Contents | 2003
|Component-Level Dataflow Analysis
British Library Conference Proceedings | 2005
|The Mosaic CGNS Dataflow Platform
AIAA | 2019
|