LLM-Assisted RTL Generation and PPA Optimization for Domain-Specific Hardware Accelerators

Authors

  • Azahar Bernad Department of Computer Science, University of New Hampshire, Durham, NH, USA. Author
  • Marissa Glass Department of Electrical Engineering and Computer Science, University of Missouri, Columbia, MO, USA. Author
  • Niyaga Suwarno Department of Computer Science, University of Central Florida, Orlando, FL, USA. Author

Keywords:

large language models, register transfer level, hardware accelerators, power performance area, electronic design automation, verification, domain-specific architecture, technical governance, sustainability, robustness

Abstract

Large language models are rapidly moving from general software assistance toward specialized engineering domains, and register transfer level design has become a consequential test case for their maturity. This paper examines the system-level implications of using large language models to generate and optimize register transfer level code for domain-specific hardware accelerators, with particular attention to power, performance, and area outcomes. Rather than treating model output as an isolated artifact, the analysis positions generated hardware descriptions inside a socio-technical pipeline that includes specification interpretation, architectural exploration, verification, synthesis, physical implementation, deployment, and long-term maintenance. The paper argues that the central challenge is not simply producing syntactically valid Verilog or VHDL, but producing designs that are robust across process corners, understandable to human engineers, governable under institutional review, and sustainable in energy and economic terms. It discusses the trade-offs between aggressive optimization and verification burden, the role of domain adaptation and benchmark design, the infrastructure required for traceability and reproducibility, and the policy questions raised by automated hardware generation. Case illustrations from accelerator design, edge inference, and cloud-scale deployment show that large language models can compress early-stage design effort while also introducing new forms of technical debt, security risk, and fairness concern. The paper concludes that responsible adoption requires hybrid workflows, rigorous verification, transparent model reporting, and governance mechanisms that treat hardware generation as a matter of public consequence rather than a purely private productivity gain.

References

1. Vaswani, A., Shazeer, N., Parmar, N., Uszkoreit, J., Jones, L., Gomez, A. N., Kaiser, Ł., & Polosukhin, I. (2017). Attention is all you need. Advances in Neural Information Processing Systems, 30, 5998–6008.

2. Brown, T. B., Mann, B., Ryder, N., Subbiah, M., Kaplan, J., Dhariwal, P., ... & Amodei, D. (2020). Language models are few-shot learners. Advances in Neural Information Processing Systems, 33, 1877–1901.

3. Chen, M., Tworek, J., Jun, H., Yuan, Q., Pinto, H. P. d. O., Kaplan, J., ... & Zaremba, W. (2021). Evaluating large language models trained on code. arXiv preprint arXiv:2107.03374.

4. Austin, J., Odena, A., Nye, M., Bosma, M., Michalewski, H., Dohan, D., ... & Sutton, C. (2021). Program synthesis with large language models. arXiv preprint arXiv:2108.07732.

5. Thorat, K., Zhao, J., Liu, Y., Hasan, A., Peng, H., Xie, X., ... & Ding, C. (2025, September). LLM-VeriPPA: Power, Performance, and Area Optimization aware Verilog Code Generation with Large Language Models. In 2025 ACM/IEEE 7th Symposium on Machine Learning for CAD (MLCAD) (pp. 1-7). IEEE.

6. Lu, Y., Liu, S., Zhang, Q., & Xie, Z. (2024). RTLLM: An open-source benchmark for design RTL generation with large language model. In 2024 29th Asia and South Pacific Design Automation Conference (ASP-DAC) (pp. 722–727). IEEE.

7. Liu, M., Ene, N., Kiran, R., Das, D., Shah, A., Hofmann, A., ... & Kan, E. C. (2023). ChipNeMo: Domain-adapted LLMs for chip design. arXiv preprint arXiv:2311.00176.

8. Blocklove, J., Garg, S., Karri, R., & Pearce, H. (2023). Chip-Chat: Challenges and opportunities in conversational hardware design. In 2023 ACM/IEEE 5th Workshop on Machine Learning for CAD (MLCAD) (pp. 1–6). IEEE.

9. Thakur, S., Ahmad, B., Fan, Z., Pearce, H., Tan, B., Karri, R., Dolan-Gavitt, B., & Garg, S. (2023). Benchmarking large language models for automated Verilog RTL code generation. In 2023 Design, Automation & Test in Europe Conference & Exhibition (DATE) (pp. 1–6). IEEE.

10. Liu, M., Pinckney, N., Khailany, B., & Ren, H. (2023). VerilogEval: Evaluating large language models for Verilog code generation. In 2023 IEEE/ACM International Conference on Computer-Aided Design (ICCAD) (pp. 1–8). IEEE.

11. Amodei, D., Olah, C., Steinhardt, J., Christiano, P., Schulman, J., & Mané, D. (2016). Concrete problems in AI safety. arXiv preprint arXiv:1606.06565.

12. Bender, E. M., Gebru, T., McMillan-Major, A., & Shmitchell, S. (2021). On the dangers of stochastic parrots: Can language models be too big? In Proceedings of the 2021 ACM Conference on Fairness, Accountability, and Transparency (pp. 610–623). ACM.

13. Mitchell, M., Wu, S., Zaldivar, A., Barnes, P., Vasserman, L., Hutchinson, B., ... & Gebru, T. (2019). Model cards for model reporting. In Proceedings of the Conference on Fairness, Accountability, and Transparency (pp. 220–229). ACM.

14. Sculley, D., Holt, G., Golovin, D., Davydov, E., Phillips, T., Ebner, D., ... & Dennison, D. (2015). Hidden technical debt in machine learning systems. Advances in Neural Information Processing Systems, 28, 2503–2511.

15. Dean, J., & Barroso, L. A. (2013). The tail at scale. Communications of the ACM, 56(2), 74–80.

16. Esmaeilzadeh, H., Blem, E., St. Amant, R., Sankaralingam, K., & Burger, D. (2011). Dark silicon and the end of multicore scaling. In 2011 38th Annual International Symposium on Computer Architecture (ISCA) (pp. 365–376). IEEE/ACM.

17. Hennessy, J. L., & Patterson, D. A. (2019). A new golden age for computer architecture. Communications of the ACM, 62(2), 48–60.

18. Horowitz, M. (2014). Computing's energy problem (and what we can do about it). In 2014 IEEE International Solid-State Circuits Conference Digest of Technical Papers (ISSCC) (pp. 10–14). IEEE.

19. Jouppi, N. P., Young, C., Patil, N., Patterson, D., Agrawal, G., Bajwa, R., ... & Yoon, D. H. (2017). In-datacenter performance analysis of a tensor processing unit. In 2017 ACM/IEEE 44th Annual International Symposium on Computer Architecture (ISCA) (pp. 1–12). ACM/IEEE.

20. Sze, V., Chen, Y.-H., Yang, T.-J., & Emer, J. S. (2017). Efficient processing of deep neural networks: A tutorial and survey. Proceedings of the IEEE, 105(12), 2295–2329.

Downloads

Published

2026-11-13