What is it about?
In this paper, we investigate how quantum computers can be used to solve the complex placement problem which appears in designing Field-Programmable Gate Arrays (FPGAs). We propose a new method tailored for current quantum hardware capabilities. Adiabatic quantum computing (AQC) is particularly promising for tackling these intricate problems due to its ability to explore vast solution spaces efficiently. Instead of solving the entire problem in one go, we break it down into smaller parts known as quadratic unconstrained binary optimization (QUBO) problems, which are then solved using AQC. Our approach allows for easy integration of design constraints and can adapt to the available hardware resources. We also evaluate the performance of contemporary quantum hardware, like the D-Wave Advantage 5.4 quantum annealer, and find promising results indicating its suitability for real-world FPGA placement tasks.
Featured Image
Photo by Christian Wiediger on Unsplash
Why is it important?
Placement is a vital step in programming FPGAs. The time needed for executing this step is crucial for the overall programming time which is the bottleneck in research applications for FPGAs. The quality of the placement does not only affect the time needed for the subsequent routing procedurethe but also the quality of the overall final chip design in terms of maximum clock frequency. Thus, to harness the true potential of FPGAs for cutting-edge technologies such as AI, optimizing the solution quality and time efficiency of the placement step is of great interest. We show that quantum computing may be of great value for this.
Perspectives
Read the Original
This page is a summary of: FPGA-Placement via Quantum Annealing, April 2024, ACM (Association for Computing Machinery),
DOI: 10.1145/3626202.3637619.
You can read the full text:
Contributors
The following have contributed to this page