Notice: The reproducibility variables underlying each score are classified using an automated LLM-based pipeline, validated against a manually labeled dataset. LLM-based classification introduces uncertainty and potential bias; scores should be interpreted as estimates. Full accuracy metrics and methodology are described in Coakley et alK. L. Coakley, T. Snelleman, H. Hoos, and O. E. Gundersen, "The embrace of open science: An analysis of a decade of AI research and 56 800 conference papers," Under Review, 2026..
SymRTLO: Enhancing RTL Code Optimization with LLMs and Neuron-Inspired Symbolic Reasoning
Authors: Yiting Wang, Wanghao Ye, Ping Guo, Yexiao He, Ziyao Wang, Bowei Tian, Shwai He, Guoheng Sun, Zheyu Shen, Sihan Chen, Ankur Srivastava, Qingfu Zhang, Gang Qu, Ang Li
NeurIPS 2025 | Venue PDF | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Experiments on the RTL-Rewriter benchmark with Synopsys Design Compiler and Yosys show that Sym RTLO improves power, performance, and area (PPA) by up to 43.9%, 62.5%, and 51.1%, respectively, compared to the state-of-the-art methods. Our code is available at https://github.com/Nelly W8/Sym RTLO |
| Researcher Affiliation | Academia | 1University of Maryland 2City University of Hong Kong 3University of Southern California EMAIL EMAIL |
| Pseudocode | Yes | A FSM Symbolic System The following section demonstrates an example FSM in verilog. First the verilog is transformed to Symbolic Representation, then the Symbolic system applied minimization algorithm to optimize the FSM. ... Listing 2: LLM-generated symbolic representation for example1_state. ... Listing 3: Reduced states of example1_state. |
| Open Source Code | Yes | Our code is available at https://github.com/Nelly W8/Sym RTLO |
| Open Datasets | Yes | Experiments on the RTL-Rewriter benchmark with Synopsys Design Compiler and Yosys show that Sym RTLO improves power, performance, and area (PPA) by up to 43.9%, 62.5%, and 51.1%, respectively, compared to the state-of-the-art methods. ... For a comprehensive evaluation, we analyze Sym RTLO performance using circuits from the RTLRewriter Benchmark. |
| Dataset Splits | No | No explicit training/test/validation dataset splits (e.g., percentages or sample counts) are provided for the RTLRewriter Benchmark or other datasets used in the experiments. The paper mentions evaluating on "11 short-benchmark examples from RTLRewriter" and "10 complex FSM and algorithm examples" which describes the cases used, but not how a dataset was split for training/testing. |
| Hardware Specification | No | The paper does not provide specific hardware details (like GPU/CPU models, memory amounts, or detailed computer specifications) used for running its experiments. It mentions commercial tools like "Synopsys DC Compiler 2019" and LLMs like "GPT-4o," but not the underlying physical hardware used by the authors. |
| Software Dependencies | Yes | Implementations The Sym RTLO framework takes GPT-4o as its primary LLM for optimization strategy selection, symbolic system generation, and iterative HDL synthesis, leveraging its robust inference, low inference cost, and coding capabilities. ... Pyverilog [34] is used for AST extraction and code reconstruction. ... To efficiently retrieve relevant transformation templates and knowledge, we integrate Open AI s text embedding-3-small, which excels in embedding-based retrieval tasks. For hardware compilation and validation, we use a combination of open-source and commercial tools. Yosys measures wires and cells, while Synopsys DC Compiler 2019 [33], paired with the Synopsys Standard Cell (SSC) library, performs PPA analysis. GPT-4o generates test benches for functional coverage. Yosys + ABC serves as the logical equivalence checker, while Synopsys Formality for sequential equivalence checking. |
| Experiment Setup | Yes | For a fair comparison with standard compiler workflows, we apply typical Synopsys DC Compiler optimizations, using medium mapping effort and incremental mapping to reflect common practices. |