{"ID":5937844,"CreatedAt":"2026-07-07T03:14:33.014478982Z","UpdatedAt":"2026-07-09T00:51:22.794632408Z","DeletedAt":null,"paper_url":"https://arxiv.org/abs/2607.04577","arxiv_id":"2607.04577","title":"Beyond the Need for Speed: Energy-Aware Code Generation via Simulation-Guided Reinforcement Learning","abstract":"Code models strictly prioritize functional correctness, leaving software energy efficiency as an unoptimized byproduct. Training models to generate energy-efficient code requires reproducible feedback at scale, which physical hardware measurement cannot reliably provide due to variance. In this paper, we replace hardware profiling with a deterministic architectural simulation harness to build Green Tea, a corpus of $3.5$ million evaluations across $1{,}474$ C++ problems. We train an energy-aware code model via supervised fine-tuning on energy-contrastive pairs, followed by closed-loop reinforcement learning (GRPO) using simulation-in-the-loop feedback. To rigorously evaluate deployment readiness, we introduce the Correctness-Adjusted Reduction in Energy Total (CARET), a metric that explicitly penalizes code that sacrifices functionality for efficiency. On $143$ held-out problems, our simulation-in-the-loop pipeline achieves $12.63\\%$ CARET, nearly tripling the gain of fine-tuning alone, and successfully beats the energy efficiency of human-expert references on $58.4\\%$ of its valid outputs. Furthermore, our analysis exposes the IPC trap: standard throughput proxies like Instructions-Per-Cycle (IPC) actively misrank true energy efficiency on $67.8\\%$ of problems, proving the absolute necessity of direct energy simulation. By releasing our dataset and infrastructure, we bypass the $263{,}000$ CPU-hours required for reproduction, structurally empowering the community to deploy inherently energy-efficient code generation models.","short_abstract":"Code models strictly prioritize functional correctness, leaving software energy efficiency as an unoptimized byproduct. Training models to generate energy-efficient code requires reproducible feedback at scale, which physical hardware measurement cannot reliably provide due to variance. In this paper, we replace hardwa...","url_abs":"https://arxiv.org/abs/2607.04577","url_pdf":"https://arxiv.org/pdf/2607.04577v1","authors":"[\"Saurabhsingh Rajput\",\"Tushar Sharma\"]","published":"2026-07-06T01:04:40Z","proceeding":"cs.LG","tasks":"[\"cs.LG\",\"cs.SE\"]","methods":"[\"Reinforcement Learning\"]","has_code":false}