RTL to GDSII: Design and Physical Implementation of a 32-bit RISC-V Floating-Point Co-Processor Using Verilog and Cadence Tools
The paper presents the RTL-to-GDSII implementation of a 32-bit RISC-V floating-point co-processor in Verilog and standard Cadence tools. The co-processor adheres to IEEE 754 on all single-precision arithmetic operations to speed up the processing of floating-point calculations for embedded systems. The design methodology includes architectural modelling, RTL design, synthesis, place-and-route (PnR), and tapeout level sign-off verification. A systematic methodology was pursued for achieving optimal trade-off between performance, area, and power. The final implementation is realized in 90 nmCMOS technology, achieving a maximum frequency of 210 MHz and consuming 38.5 mW in normal conditions. This shows the effectiveness of merging the RISC-V modular architecture with floating-point capabilities in a low-power domain.
Authors
- Chandrashekar M Patil (ORCID: https://orcid.org/0000-0002-5220-4799)
- Ramashankar Yadav
- Shashank M. Dravid (ORCID: https://orcid.org/0000-0002-9912-8148)
- Ravi N. Srinivasa (ORCID: https://orcid.org/0000-0003-0984-3664)
- A Geethashree
- S Ajay H
Institutions
- ASA College (US)
Publication Details
- Published
- 2025-06-20
- DOI
- https://doi.org/10.1109/conit65521.2025.11167812
- Primary Topic
- Embedded Systems Design Techniques
- Type
- article
- Field-Weighted Citation Impact
- 0.00