COEVO: Co-Evolutionary Framework for Joint Functional Correctness and PPA Optimization in LLM-Based RTL Generation CCS: Computing methodologies Artificial intelligence CCS: Hardware Hardware description languages and compilation CCS: Hardware High-level and register-transfer level synthesis

Heng Ping, Peiyu Zhang, Shixuan Li, Wei Yang, Anzhe Cheng, Shukai Duan, Xiaole Zhang, Paul Bogdan · 2026

Read COEVO: Co-Evolutionary Framework for Joint Functional Correctness and PPA Optimization in LLM-Based RTL Generation CCS: Computing methodologies Artificial intelligence CCS: Hardware Hardware description languages and compilation CCS: Hardware High-level and register-transfer level synthesis by Heng Ping, Peiyu Zhang, Shixuan Li, Wei Yang, Anzhe Cheng, Shukai Duan, Xiaole Zhang, Paul Bogdan (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2604-15001.txt.

Description

Ping and seven others propose COEVO, unifying correctness and PPA in LLM-based RTL generation. Existing methods split correctness and PPA, delaying PPA until correctness, and discard promising partial candidates. COEVO treats correctness alongside area, delay, and power, with an enhanced testbench. An adaptive gate lets partial candidates guide search, and Pareto sorting keeps trade-offs without manual weights. On VerilogEval, COEVO reaches 97.5 percent Pass@1 with GPT-5.4-mini, beating baselines and attaining best PPA on 43 of 49 RTLLM designs.

Catalog record

CREATORHeng Ping, Peiyu Zhang, Shixuan Li, Wei Yang, Anzhe Cheng, Shukai Duan, Xiaole Zhang, Paul Bogdan
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE47178
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2604.15001

Browse the full library · About the Tower of Babel