SynAct: A Reasoning-Acting Large Language Model Agent for Adaptive Synthesis Optimization

ARES: Adaptive Reasoning-Effort Steering for PPA- and Cost-Aware RTL Optimization with LLM Agents

Large language model (LLM) agents optimize the power, performance, and area (PPA) of register-transfer-level (RTL) designs by iterating over edits, synthesis, and PPA analysis, paying a dollar cost for every LLM call. Prior agents report the quality reached without its normalized cost, attribute that quality to an engineered cross-design memory, and hold the reasoning effort of every call fixed. We propose Ares with three corresponding innovations. (1) We introduce a normalized dollar cost per LLM call reported alongside the figure of merit (FoM), enabling fair comparison across effort levels and optimizers. (2) Using this accounting, we find the construction of the long-term memory matters little. An engineered memory brings no dependable gain over a plain concatenation of the same experience. (3) We instead adapt the per-call reasoning effort by escalating to deeper reasoning only once progress at a lower effort stalls, via a patience counter fit on 21 training designs, allocating reasoning where it pays rather than uniformly across all iterations. On three test designs unseen during training, the effort policy lowers the FoM by 23-27% where the best fixed effort reaches 16-23%, at equal normalized cost. Ares closes up to 83% of the gap from an LLM-drafted multiply-accumulate unit to its highly hand-optimized counterpart, and reaches a 25% deeper FoM than state-of-the-art Dr. RTL at 12% of its tokens.

Semantic Mutation via LLMs: A Hybrid Approach to Evolutionary Program Synthesis

This work proposes a hybrid framework that integrates Large Language Models into GP in two complementary roles: as a semantic mutation operator that proposes non-local program rewrites during evolution, and as a post-evolution repair that iteratively refines single failed programs after search terminates.

SeqFeed: Improving Agentic RTL Code Generation with Sequential Behavior Feedback

RTL code generation is a critical stage in hardware design, and the emergence of agentic systems offers new opportunities to automate this process. To generate correct RTL code, agents must understand sequential behavior, including how signals evolve and propagate over multiple clock cycles. However, effectively conveying such temporal information to agents remains a significant challenge. RTL code does not expose cycle-level signal behavior for a specific execution, whereas full simulation waveforms are too voluminous and noisy for effective LLM analysis. To address these limitations, we study how human engineers reason about sequential behavior and identify three requirements for effective feedback: it should be event-addressable, dependency-traceable, and iteratively-queryable. Guided by these requirements, we propose \textit{SeqFeed}, which comprises two complementary mechanisms: (1) \textit{SeQuery}, an SQL-like waveform query language that enables agents to anchor queries to semantic events and sample signal values at relative time points; and (2) \textit{SeGraph}, a dependency graph that tracks signal propagation across clock cycles. Experimental results across multiple LLMs demonstrate the effectiveness of SeqFeed in improving pass rates. SeQuery and SeGraph are each effective independently and provide complementary benefits when used together.

Verifiable PDE Reasoning and Modeling with Neurosymbolics

Timing-driven Approximate Logic Synthesis Based on Cooperative Search

Approximate logic synthesis (ALS) is a promising design paradigm for error-resilient applications. It can automatically generate circuits with optimized delay, area, and power at the cost of small errors. Existing methods mainly apply local approximate changes (LACs) to shorten critical paths or reduce area. However, given the large search space of LACs, they struggle to achieve a good trade-off between both objectives. This restricts the timing optimization potential of approximate circuits, particularly since area savings can be reinvested to enhance the drive strength of timing-critical gates. This paper leverages population-based search to explore LACs with positive timing influence in parallel. We develop two LAC search actions specialized in critical path shortening and area reduction, respectively. Guided by a recommender system aware of circuit topology and quality, individuals iteratively execute suitable LAC search actions and refine their explored LAC sets by conflict-free combinations, thus balancing reductions in critical path depth and area. Area savings are then reinvested to enhance the drive strength of timing-critical gates, enabling further timing improvement. Experimental results show that compared to a state-of-the-art area-driven method, our framework further reduces 20.3% critical path delay with comparable area savings, while being 7.9× faster.

From Atomic Actions to Standard Operating Procedures: Iterative Tool Optimization for Self-Evolving LLM Agents

EvoSOP is introduced, a framework that empowers agents to extract SOPs from execution trajectories and iteratively optimize the toolset through a systematic lifecycle of construction, merging, evaluation, and pruning, providing a scalable pathway for the development of self-evolving agents.