Code Optimization
Latest papers
26 papers4 papers in the last four weeks, with none the four weeks before. 0.0% of all new papers.
Program optimization aims to improve performance while preserving behavior. Coding agents can already make programs faster by changing algorithms, data structures, and execution strategies. However, existing equivalence checks mostly rely on tests, which agents have been shown to exploit, passing the tests without preserving the intended behavior. We propose agentic verified optimization, in which an agent produces a faster implementation together with a machine-checked proof of equivalence to the existing one. Because the existing implementation serves as the specification, the task avoids the specification-writing bottleneck of verified code generation, and recent advances in theorem proving make the proofs attainable. We formulate the task at the function and repository levels in Lean 4, where programs can be both executed and reasoned about. Few real-world applications are written in Lean, so we develop Aqueduct, an agentic framework that migrates C, C++, Go, and Rust programs into native Lean, validates each port by differential testing against its original. On the 189 reference implementations in Verina, agents produce faster proved rewrites for 156, including algorithmic changes such as replacing exponential recursion with dynamic programming. On 17 migrated applications, their 92 proved rewrites speed up 47 of 51 application scenarios. These application-level gains are smaller than the function-level ones, making the choice of functions to optimize an important problem. Current agents can thus already carry out verified optimization on both standalone functions and applications migrated from real programs. We discuss the remaining challenges and the steps toward verified optimization in production.
FrontierRefactor: Benchmarking Multi-Scale Performance-Oriented Codebase Refactoring
Autonomous codebase refactoring requires coding agents to restructure working software while strictly preserving externally observable behavior and prioritizing execution performance and compute efficiency over stylistic preference. Existing benchmarks: refactoring benchmarks assess functional correctness while overlooking runtime regressions, whereas code performance benchmarks target isolated micro-kernels that ignore codebase-scale complexity. We observe that refactoring scale is governed by the scope of the observable contract, the caller-visible interface fixing behavioral invariants while freeing private implementations for optimization. Building on this insight, we introduce FrontierRefactor, a benchmark of 60 qualified engineering tasks spanning four scales, from computational kernels to full repositories, across in-place optimization and cross-language rewriting. Evaluating frontier coding agents through a reference-free cascaded protocol across 1,080 execution runs reveals that correctness and speedup are largely decoupled: 41.2% of submissions satisfy functional equivalence yet fail to deliver verified performance improvements, with 19.3% suffering performance regressions and 21.9% trapped in measurement noise. Refactoring paradigms reveal orthogonal engineering bottlenecks: while cross-language rewrites gain speedups from native compilation after passing functional checks, agents struggle with contract fidelity, with pass rates dropping from 65.0% on modules to 32.6% at package scale; conversely, in-place optimization reliably preserves behavioral contracts at 86.5% but rarely achieves acceleration without physical performance profiling. Trajectory analysis shows that agents fall victim to uncalibrated micro-benchmarking illusions, an operational blind spot that multi-turn execution-timing feedback directly resolves, raising strict success from 16.0% to 40.0%.
Where Does the Performance Go? Understanding and Recovering Performance Regressions in C-to-Rust Translation
Rust is increasingly adopted as a memory-safe successor to C, and transpilers such as \crust{} can migrate entire C projects while preserving their behavior. Preserving behavior, however, does not guarantee preserving performance. We study this gap on 20 real-world C projects and their \crust{} translations. On the eight projects used for analysis, the translated Rust is more than 3% slower than the original C on 12 of 19 operations, by up to 62.2%. By comparing execution-cost profiles, optimized LLVM IR, and compiler optimization remarks across each C--Rust pair, we trace these slowdowns to nine recurring performance issues, spanning residual runtime work, missed compiler optimizations, and C-style source patterns, and derive an optimization direction for each. We then present \ourtool{}, an LLM-based approach that encodes these issues as reusable optimization rules to recover the performance of translated Rust. \ourtool{} profiles each project to locate hot functions and matches them against rule-specific patterns. The LLM then checks each rule's rewrite preconditions against the project context before rewriting, and a rewrite is kept only if it passes both functional and performance validation. On 41 operations from twelve held-out projects, \ourtool{} preserves C outputs on all tested workloads and speeds up 40 operations over the \crust{} translations, 18 of which end up faster than the original C. It outperforms the general-purpose coding agent \codex{} on 31 of the 37 operations where both preserve C outputs, and outperforms its own variant without rule guidance on 34 of the 41 operations.
How to Train Your Little Dragon: Teaching a Small Language Model to Construct and Refine GPU Optimization Scripts
Compiler transformations interact, making GPU kernel optimization a search for combinations that preserve the computation while improving performance. We introduce HOW TO TRAIN YOUR LITTLE DRAGON, a training method that teaches a small language model both to construct complete optimization scripts and to improve existing ones. Instead of ranking candidates or generating kernel code, our model generates optimization scripts. We devise a compact, lossless representation whose deterministic decoding reconstructs the exact compiler trace, preserving its transformation parameters and dependencies. We use a compiler framework TVM MetaSchedule to build training data on 54,116 distinct workloads, and construct 108,594 examples in two training sets: 60,236 teach the model to generate an optimized script from an unoptimized kernel, while 48,358 teach it to produce a better script given the kernel and a somewhat optimized script. Supervised fine-tuning teaches the model to construct and improve optimization scripts from these search-derived examples. We then add reinforcement learning with a shaped reward that combines correctness with measured performance, awarding a latency-based bonus only to numerically correct kernels. We apply this training method to a model with 8 billion parameters. After supervised fine-tuning and reinforcement learning, the model achieves per-attempt correctness of 91.27% for construction and 92.01% for refinement, while reinforcement learning produces faster kernels than supervised training on 88 of 102 construction tasks and 145 of 170 refinement prompts. With only 8B parameters, our LITTLE DRAGON achieves higher per-attempt correctness and better kernel performance than six frontier LLMs: GPT-6 Astra, Opus 5, Kimi-K3, and GLM-5.3 among them, each with hundreds of billions to trillions of parameters, for which task-specific training is unavailable or expensive.
Efficiency Hallucination: Formalizing and Measuring Behavioral Calibration in LLM-Based Code Optimization
The integration of Large Language Models (LLMs) into automated code optimization introduces a critical reliability risk we term the Efficiency Hallucination: an LLM's tendency to issue non-functional mutations with unsubstantiated performance claims on already-optimized code. This is driven by the Evaluation Trap, wherein binary benchmarks incentivize unnecessary modifications over safely abstaining. We present a validation framework using classification penalty methods, evaluated across 180 optimization runs on nine models (GPT, Claude, Gemini) using EffiBench. Under standard prompts, models exhibit a 100% over-edit rate on optimal code. Our guardrail raises correct abstention from 0% to to 44.4%, preserving a 100% edit rate on sub-optimal code with zero false abstentions. Calibration is uneven: GPT-5.4 Mini approaches near-perfect abstention, and simple code is recognized more reliably than complex code. Our framework offers a training-free mechanism to mitigate LLM overconfidence before deployment in production.
Loreley: Repository-Scale Program Evolution with Quality-Diversity Search
Sequential agent search accumulates changes from its current champion but discards alternative branches; independent proposals preserve breadth but restart from the root. Loreley instead retains complete repository states in a Quality-Diversity (QD) archive and samples them as parents or supplies them as context for later edits. Candidates are Git commits produced in isolated worktrees and judged by a project-supplied evaluator. We compare configured Loreley QD, sequential champion editing, and independent root proposals in a matched Zstandard experiment: seven paired blocks and 48 physical candidate jobs per policy and block (1,008 total), with root-only initialization and each policy's native concurrency. Validation selected a winner at each budget checkpoint; an agent-hidden holdout measured the fixed candidate. At 48 jobs, QD was 0.135% below Sequential Champion (95% BCa interval for the paired effect: -0.556% to +0.161%) and 0.320% above Independent Root (-0.082% to +0.686%). Neither contrast established a QD advantage; Sequential had the highest observed 48-job mean and median. Archive retention and later sampling did occur. Four of seven final QD winners had a non-incumbent state in their primary-parent ancestry under a retrospective one-incumbent rule applied only to the observed QD stream. Including inspiration edges raised the count to six, without showing that supplied context caused an edit. Three earlier capability campaigns produced generation-4, multi-file improvements in two Python libraries and a separate Zstandard revision. Loreley engaged the intended stepping-stone mechanism, but the controlled experiment did not show an endpoint benefit at 48 jobs.
T-LLM Compiler: Trusted LLM-based Code Optimization and Verification Framework
Recent advances in Large Language Models (LLMs) have opened opportunities to apply high-level code transformations to the field of code optimization, and it has since emerged as one of the most fundamental tasks for LLMs to perform; however, at present, LLMs struggle to apply wide-ranging code optimization tasks due to both the complexity of the code and the inability to independently verify the correctness of the transformations. In this paper, we present the Trusted LLM (T-LLM) Compiler, which proposes an advancement in compiler technology through a collaborative effort involving high-level LLM code transformations, traditional compilers, and verification tools. Experimental results reveal that it can significantly improve code correctness when tested on a set of PolyBench/C benchmarks. Our approach facilitates iterative code optimization efforts with verification strategies that enable corrective actions. Through this approach, T-LLM Compiler achieves code optimization accuracy of up to 83.3% and a speedup of up to 16.1% on the PolyBench/C benchmarks, with the transformed code reaching an average of 26.7% speedup wrt standard baselines. Additionally, we release the project's source code to the open-source community.
Effect of Abstractions and Prompting Strategies on LLM-Guided High-Performance Optimizations
Code performance optimization is a vital aspect of modern software development, as it enables faster response times and reduced resource usage. These optimizations require a deep understanding of low-level hardware details and the intricacies of parallel processing, making them challenging even for experienced developers. With the advent of Large Language Models (LLMs), which are increasingly capable of generating and understanding code, there is growing interest in incorporating these models into automated code optimization processes. Traditionally, this automation involves transcribing the source code into a domain-specific representation that can be auto-tuned using grid search or machine learning algorithms, while adhering to strict rules and a limited set of feasible transformations to ensure verifiability. LLMs incorporate high-level code semantics and can thus perform transformations that go beyond verifiable automated optimizations. This paper investigates whether the traditional abstractions used in automated code optimization improve the performance and correctness of LLM-guided optimizations of parallel HPC applications. We evaluate this using the PolyBench benchmark suite and demonstrate that, in our evaluated setting, LLMs provided with specific optimization goals achieve better measured performance and validity rates when generating C code compared to creating computation pipelines and optimization schedules with established frameworks, suggesting that future development should explore alternative approaches for verifiable LLM-guided code optimization.
RLPF: Reinforcement Learning from Performance Feedback for Code Generation
Code models are increasingly trained with execution feedback, but most training signals still stop at correctness. This leaves an important gap for systems code: two programs can pass the same tests while differing greatly in runtime. We study how to train code agents to prefer faster correct implementations, rather than treating efficiency only as an evaluation metric. The key difficulty is that runtime is a fragile reward. It is meaningful only after a program is correct, varies across tasks, and gives little guidance when most sampled programs fail to compile or run. We propose RLPF, reinforcement learning from performance feedback, which turns execution outcomes into a staged reward. Failed programs are ordered by execution progress, while correct programs are ranked by their relative improvement from the baseline toward the expert reference. This gives useful feedback before correctness and performance-sensitive feedback after correctness. Fine-tuning Qwen3-32B with RLPF on PerfCodeBench raises correct-and-runnable solutions from to and improves relative efficiency from to . The trained model becomes competitive with stronger open-weight systems, and its optimization behavior transfers modestly to EffiBench-X. Additional studies show that model-generated references provide useful but weaker supervision, and that the full composite reward is more reliable than correctness-only or runtime-only baselines. These results suggest that code agents can be trained not only to pass tests, but also to optimize the programs they write.
Reinforcement Learning for Code Optimization
RL for code correctness is now established: have the model generate a program, run it against hidden test cases, and reward solutions that pass. Extending this to code optimization seems straightforward: just add execution time to the reward. But in practice, once timing drives the reward, small problems in measurement noise, reward sparsity, or GRPO instability overwhelm the signal and make RL fail: generated solutions are barely faster, and more of them can fail. We make execution time learnable through three stages: (1) how code is tested, by building DMC-Optim with large optimization tests and a calibrated sandbox; (2) how speed is turned into reward, by composing correctness and speed in the RL environment and using an offline simulator to predict the most promising configurations; and (3) how the model learns from that reward, by adapting GRPO and evaluation to the sparser, noisier timed-execution setting. On DMC-Optim, the strongest optimization-aware configurations improve strict top-50% pass@1 from 18.0% to 31.3% on Qwen 2.5 7B and from 30.7% to 50.4% on CWM 32B. These gains further increase at stricter percentiles such as top-30%, with 125% relative improvement for CWM 32B, while preserving pure-correctness scores. When the timing sandbox is degraded, robust optimization RL reaches 100% to 200% improvement over standard RLVR, depending on the evaluation criterion. On LCB, CWM 32B wins up to 83% of median-sample speed comparisons against standard RLVR. Relative to the fastest correct human submissions per problem, it reaches about half the human rate of complexity-class improvements (13% vs. 22%).
Teaching LLMs to Self-Evolve: Cultivating Core Meta-Skills with Reinforcement Learning
Test-time scaling through iterative self-evolution with environment feedback, as demonstrated by AlphaEvolve, shows remarkable performance gains. We hypothesize that the success of such evolution frameworks hinges on meta-skills, such as self-reflection with environment feedback, that enable effective multi-round refinement, yet are largely neglected by traditional post-training. To bridge this gap, we present MetaEvolve, a framework designed to develop these meta-skills via a data synthesis pipeline, evolution-aware reinforcement learning (RL), and inference-time evolutionary search. Concretely, we ground MetaEvolve in coding, where program execution provides natural, continuous reward signals beyond binary correctness. Building on these signals, we synthesize evolution trajectories as training data, each containing a current program, its fitness score (combining correctness and efficiency), and a history of prior attempts, and train the model via RL with verifiable rewards derived from test case execution. By training on large-scale code data, we aim to inspire generalizable domain-agnostic meta-skills that can transfer broadly to open-ended problems where such rich training signals are scarce. Across seven coding benchmarks, MetaEvolve outperforms the strongest baseline by 10.01% absolute on in-distribution tasks and 24.12% on out-of-distribution tasks. On open-ended algorithm optimization problems entirely outside the training domain, it further achieves a 46.9% relative improvement. These results demonstrate that explicitly cultivating self-evolution meta-skills offers a principled path toward more capable and autonomously self-evolving AI.
PerfAgent: Profiler-Guided Iterative Refinement for Repository-Level Code Optimization
Large language model (LLM) agents now perform well on correctness-oriented repository-level tasks, including SWE-Bench issue resolution and feature implementation in real codebases. However, they still struggle with repository-level code optimization, which requires preserving behavior while improving runtime performance. Passing tests is not enough in this setting; a patch must preserve behavior, implement code optimization, and approach expert speedups. Current agents often miss bottlenecks hidden behind abstraction layers and native extensions, stop after shallow speedups, or insufficiently test the code patches that thus may silently break edge cases. We present PerfAgent, a profiler-guided, verifier-in-the-loop workflow that gives an off-the-shelf coding agent the feedback needed to find real hotspots, improve beyond the first passing patch, and use profiler evidence rather than timing alone to decide what to optimize next. On two challenging optimization benchmarks, GSO and SWE-fficiency-Lite, PerfAgent more than doubles the rate of expert-matching patches over OpenHands with GPT-5.1, improving from 19.6% to 39.2% on GSO and from 26% to 74% on SWE-fficiency-Lite. It also surpasses an oracle best-of-five baseline at substantially lower cost, showing that the gains come from better feedback rather than additional test-time sampling.
Understanding Agent-Based Patching of Compiler Missed Optimizations
Compiler missed optimizations refer to cases in which compilers failed to optimize certain code. It takes many compiler developers' efforts to implement or patch such missed optimizations. In this paper, we present a systematic study of how well agents patch compiler missed optimizations. We identify a significant challenge that patching a missed optimization requires more than just fixing the reported case, and instead requires generalizing to similar cases. We construct a benchmark of real-world LLVM missed optimization issues and compare agent-generated patches with patches from developers in terms of optimization scope. Our results show that coding agents often optimize the given examples, but many generated patches either cover only part of the developer-intended scope or partially overlap with it; in some cases, they further generalize beyond the reference patch. We further introduce historical-knowledge augmentation techniques that leverage prior LLVM optimization pull requests through retrieval and distillation, showing that they improve developer-aligned generalization and yield practical benefits when applied to real-world IR.
SWE-Opt: A Benchmark for Evaluating LLMs on Performance-Oriented Repository-Level Optimization
Software performance optimization is a notoriously complex and manual task. Despite the growing use of Large Language Models (LLMs) for code refinement, we still lack benchmarks that capture how optimization actually happens in real-world codebases. Existing frameworks often oversimplify the problem by focusing on isolated functions or a single performance metric, missing the critical trade-offs between execution time and memory footprint, the inherent noise of the measurement environment, and the variability introduced by different input data and execution conditions. We address this by introducing SWE-Opt, a repository-level benchmark derived from 102 expert-written optimizations from open-source projects. Unlike previous benchmarks, SWE-Opt pairs each task with parameterized tests to evaluate runtime, peak memory, and Time-Weighted Memory Usage (TWMU) across varying input data and execution conditions under noise-aware measurement conditions. Our evaluation shows that current LLMs struggle significantly: runtime gains are sparse and inconsistent, and memory optimizations are nearly non-existent. An agentic setting with iterative repository exploration improves patch correctness substantially but does not translate into performance gains. This stands in sharp contrast to expert implementations, which achieve an aggregate speedup of 15.5x and peak memory reduction of 171.3x over benchmark tasks. Expert-written improvements are observed in 93.1% of tasks for runtime and 70.6% for peak memory. Our findings expose a substantial gap between current LLM capabilities and the demands of expert-level engineering.
AutoPass: Evidence-Guided LLM Agents for Compiler Performance Tuning
Large Language Models (LLMs) show promise for code compilation tasks, but applying them to runtime performance tuning is difficult due to complex microarchitectural effects and noisy runtime measurements. We present AutoPass, a multi-agent framework for compiler performance tuning that uses compiler and runtime evidence to guide LLM-generated optimization decisions. Rather than treating the compiler as a black box like prior auto-tuning schemes, AutoPass opens up the compiler to the LLM, enabling it to query compiler-internal optimization states and analyze the intermediate representation to orchestrate compiler options. The search process iteratively refines optimization configurations using measured runtime feedback to diagnose regressions and guide latency-improving edits. AutoPass operates in an inference-only, training-free setting and requires no offline training or task-specific fine-tuning, making it readily applicable to new benchmarks and platforms. We implement AutoPass on the LLVM compiler and evaluate it on server-grade x86-64 and embedded ARM64 systems. AutoPass outperforms expert-tuned heuristics and classical autotuning methods, achieving geometric-mean speedups of 1.043x and 1.117x over LLVM -O3 on x86-64 and ARM64, respectively.
AI-PROPELLER: Warehouse-Scale Interprocedural Code Layout Optimization with AlphaEvolve
Post-link optimizers (PLOs) such as Propeller and BOLT have demonstrated that precise, profile-guided code layout can extract significant performance gains from heavily optimized binaries. However, these systems are currently restricted to intraprocedural techniques, leaving the global potential of interprocedural layout largely untapped. Interprocedural code layout is historically difficult due to a combinatorially intractable search space and complex call-return semantics that are challenging to model. Consequently, the performance potential of fine-grained interprocedural layout remains unproven in practice. AI-PROPELLER uses Magellan, an agentic workflow that evolves the compiler heuristic in Propeller into a fine-grained interprocedural optimizer and fine-tunes the resulting policy hyperparameters. To ensure high-fidelity, we move away from approximate static cost models and the agentic workflow generates multiple layout variants that are executed on actual hardware to measure real performance counters, providing a precise reward signal for the evolutionary loop. AI-PROPELLER has been evaluated on several benchmarks including large warehouse-scale applications and experiments show performance improvements of 0.23% to 1.6% optimized with state-of-the-art FDO and PLO which is significant for real-world binaries. This is the first time ever that large warehouse-scale applications in industrial settings have been optimized with fine-grained interprocedural code layout.
KLineage: Recovering the Missing When of Kernel Optimization by Deoptimizing Experts
LLM-based agents are increasingly used to generate GPU kernels, but they often struggle to determine when an optimization is sound because its required code state and dependencies are implicit in expert implementations. We introduce KLineage, which learns this missing "when" knowledge from expert kernels: instead of relying on forward rollouts, KLineage walks expert implementations backward through validation-gated simplifications and reverses each accepted step into a reusable optimization skill. Each skill records not only the optimization intent, but also when to apply the optimization technique, including where it applies in code, what conditions made it valid, what effect it has, and what failures its assumptions avoid. A downstream LLM materializes these skills on new code surfaces under the same compile/correctness/profile gate. This guidance on when to apply each optimization can help downstream models to generate higher-performance kernels. On five expert workloads across two NVIDIA architectures, these lineage-derived skills serve as an effective optimization curriculum, exceeding recent memory-based LLM-kernel baselines in both final kernel quality and optimization efficiency under the same fixed budget. We also demonstrate that the KLineage framework extends beyond NVIDIA GPUs to Ascend NPUs. Our code is publicly available at https://github.com/ict-agent/klineage.
Prior Knowledge or Search? A Study of LLM Agents in Hardware-Aware Code Optimization
LLM discovery and optimization systems are increasingly applied across domains, implementing a common propose-evaluate-revise loop. Such optimization or discovery progresses via context conditioning on received feedback from an environment. However, as modern LLM agents are increasingly complex in their structure, it is difficult to evaluate which components contribute the most, and when and how this exploration may fail. We answer these questions through three controlled experiments. Our findings: (1) In pure black-box optimization, LLMs act as greedy optimizers. (2) In zero-shot kernel generation, providing explicit input-size information has no measurable effect, models converge to the same kernel parameters regardless of size or temperature, as though the size instruction were invisible. Moreover, when tasked to perform kernel optimization for uncommon kernel sizes, performance sharply degrades regardless of the language used. (3) In feedback-loop kernel optimization, CUDA improves monotonically under iterative feedback, while TVM IR actively degrades, which demonstrates that kernel optimization degrades when models operate with low-density language. Our results conclude that LLMs in code optimization tasks highly depend on pretrained priors rather than provided feedback or agentic structure.
CodeEvolve: LLM-Driven Evolutionary Optimization with Runtime-Enriched Target Selection for Multi-Language Code Enhancement
We present CodeEvolve, an evolutionary framework for improving program performance and code quality with Large Language Models (LLMs). CodeEvolve extends OpenEvolve with runtime-guided target selection, Monte Carlo Tree Search (MCTS), automated code refinement, and language-specific evaluation pipelines for Java and Salesforce Apex. The system uses Java Flight Recorder (JFR) profiles to build weighted component graphs and select optimization targets that account for most execution cost, reducing reliance on manual bottleneck identification. For each target, CodeEvolve generates candidate edits, evaluates them through build validation, unit tests, performance checks, static analysis, and LLM-based review, and retains only variants that preserve functional correctness. Across real-world optimization tasks, CodeEvolve improves performance and code metrics while maintaining correctness. On a large enterprise Java codebase, it achieves an average speedup of 15.22 across seven hotspot functions and outperforms single-pass LLM optimization on five of them. An ablation study on Apex optimization shows that the full MCTS-augmented configuration produces 19.5 valid programs out of 20 on average, indicating that search, filtering, and refinement each contribute to more reliable optimization.
EffiPair: Improving the Efficiency of LLM-generated Code with Differential Execution Feedback
Large language models (LLMs) can generate functionally correct programs that differ substantially in execution efficiency. Existing inference-time optimization methods typically refine each candidate using pointwise runtime or profiling feedback, which identifies how costly an implementation is or where the cost arises, but offers limited guidance on how the computation should change. We introduce DIFFERENTIAL EXECUTION FEEDBACK (DEF), which compares implementations that are nearby in structural program space but separated in performance, turning their execution and implementation differences into directional optimization evidence. We demonstrate DEF in EffiPair, a training-free, test-time framework that pairs structurally similar programs with different efficiencies, distills their relative execution behavior into compact feedback, and iteratively refines a candidate pool. Across EvalPerf, Mercury, and ENAMEL, using GPT-4o mini, DeepSeek-V4.1 Flash, and GPT-5 mini, EffiPair achieves the highest value on each benchmark's official efficiency metric in all nine model-benchmark settings under matched evaluation conditions. Moreover, two contrastive refinement rounds improve efficiency over the selected initial draft in every setting while preserving or improving Pass@1. These results demonstrate the effectiveness of relational execution feedback as a lightweight signal for test-time code optimization.
EffiSkill: Agent Skill Based Automated Code Efficiency Optimization
Code efficiency is a fundamental aspect of software quality, yet how to harness large language models (LLMs) to optimize programs remains challenging. Prior approaches have sought for one-shot rewriting, retrieved exemplars, or prompt-based search, but they do not explicitly distill reusable optimization knowledge, which limits generalization beyond individual instances. In this paper, we present EffiSkill, a framework for code-efficiency optimization that builds a portable optimization toolbox for LLM-based agents. The key idea is to model recurring slow-to-fast transformations as reusable agent skills that capture both concrete transformation mechanisms and higher-level optimization strategies. EffiSkill adopts a two-stage design: Stage I mines Operator and Meta Skills from large-scale slow/fast program pairs to build a skill library; Stage II applies this library to unseen programs through execution-free diagnosis, skill retrieval, plan composition, and candidate generation, without runtime feedback. Results on EffiBench-X show that EffiSkill achieves higher optimization success rates, improving over the strongest baseline by 3.69 to 12.52 percentage points across model and language settings. These findings suggest that mechanism-level skill reuse provides a useful foundation for execution-free code optimization, and that the resulting skill library can serve as a reusable resource for broader agent workflows.
FormulaCode: Evaluating Agentic Optimization on Large Codebases
Large language model (LLM) coding agents increasingly operate at the repository level, motivating benchmarks that evaluate their ability to optimize entire codebases under realistic constraints. Existing code benchmarks largely rely on synthetic tasks, binary correctness signals, or single-objective evaluation, limiting their ability to assess holistic optimization behavior. We introduce FormulaCode, a benchmark for evaluating agentic optimization on large, real-world codebases with fine-grained, multi-objective performance metrics. FormulaCode comprises 957 performance bottlenecks mined from scientific Python repositories on GitHub, each paired with expert-authored patches and, on average, 264.6 community-maintained performance workloads per task, enabling the holistic ability of LLM agents to optimize codebases under realistic correctness and performance constraints. Our evaluations reveal that repository-scale, multi-objective optimization remains a major challenge for frontier LLM agents. Project website at: https://formula-code.github.io
SemRep: Generative Code Representation Learning with Code Transformations
Code transformation is a foundational capability in the software development process, where its effectiveness relies on constructing a high-quality code representation to characterize the input code semantics and guide the transformation. Existing approaches treat code transformation as an end-to-end learning task, leaving the construction of the representation needed for semantic reasoning implicit in model weights or relying on rigid compiler-level abstractions. We present SemRep, a framework that improves code transformation through generative code representation learning. Our key insight is to employ the semantics-preserving transformations as the intermediate representation, which serves as both a generative mid-training task and the guidance for subsequent instruction-specific code transformations. Across general code editing and optimization tasks (e.g., GPU kernel optimization), SemRep outperforms the extensively finetuned baselines with strictly the same training budget by 6.9% in correctness, 1.1x in performance, 13.9% in generalization, and 6.7% in robustness. With the improved exploration of diverse code transformations, SemRep is particularly amenable to evolutionary search. Combined with an evolutionary coding agent, SemRep finds optimizations that 685B larger-weight baselines fail to discover while achieving the same performance with 25% less inference compute.
ISO-Bench: Can Coding Agents Optimize Real-World Inference Workloads?
We introduce ISO-Bench, a benchmark for coding agents to test their capabilities on real-world inference optimization tasks. These tasks were taken from vLLM and SGLang, two of the most popular LLM serving frameworks. Each task provides an agent with a codebase and bottleneck description, whereby the agent must produce an optimization patch evaluated against expert human solutions. We curated 54 tasks from merged pull requests with measurable performance improvements. While existing benchmarks heavily use runtime-based metrics, such approaches can be gamed to pass tests without capturing the actual intent of the code changes. Therefore, we combine both hard (execution-based) and soft (LLM-based) metrics to show that both are necessary for complete evaluation. While evaluating both closed and open-source coding agents, we find no single agent dominates across codebases. Surprisingly, agents often identify correct bottlenecks but fail to execute working solutions. We also show that agents with identical underlying models differ substantially, suggesting scaffolding is as important as the model.
Beyond Pass-by-Pass Optimization: Intent-Driven IR Optimization with Large Language Models
Modern compilers optimize programs through a sequence of modular passes over intermediate representations (IR). While this pass-by-pass paradigm offers engineering benefits, it suffers from a pass coordination problem: locally beneficial transformations may block more profitable optimizations in later stages. This limitation stems from the lack of an explicit notion of optimization intent, defined as a holistic strategy for coordinating multiple transformations toward a global performance objective. Recent LLM-based approaches formulate IR optimization as an end-to-end generation task, thereby avoiding the traditional pass-by-pass structure. However, optimization intent remains implicit in these methods, forcing models to jointly infer optimization strategy and generate low-level transformations, which limits both correctness and performance. We propose IntOpt, the first intent-driven IR optimizer that explicitly separates high-level optimization intent from low-level analysis and transformation. IntOpt organizes IR optimization into three stages: intent formulation, intent refinement, and intent realization, enabling globally coordinated transformations. Experiments show that IntOpt achieves 90.5% verified correctness and 2.660x average speedup on 200-program test set, outperforming state-of-the-art LLM-based optimizers in both correctness and performance, and surpassing modern compiler with the -O3 option on 37 benchmarks with speedups of up to 272.60x.
PACEvolve: Enabling Progress-Aware Consistent Evolution
Self-evolving agents powered by Large Language Models (LLMs) have emerged as a promising direction across diverse domains, including code optimization and scientific discovery, yet their core failure modes remain underexplored. Through a comprehensive empirical study, we identify that the model's reasoning becomes anchored to the local context of current hypotheses, overemphasizing low-level details while neglecting the broader search landscape. As a result, such agents become prone to context pollution and mode collapse, repeatedly revisiting flawed hypotheses and converging on suboptimal solutions. To address this challenge, we propose Progress-Aware Consistent Evolution (PACEvolve), a systematic framework for governing agent memory and search dynamics. PACEvolve overcomes these limitations through three key techniques: (1) Hierarchical Context Management (HCM), which structures historical trajectories while dynamically pruning branches to preserve a high-signal memory state; (2) Momentum-Based Backtracking (MBB), which monitors optimization progress to escape local minima; and (3) a self-adaptive Collaborative Evolution policy (CE) that balances intra-trajectory refinement with inter-trajectory knowledge transfer. By decoupling high-level idea generation from low-level code evaluation, PACEvolve maintains a global view of search momentum and achieves state-of-the-art results across complex evolutionary benchmarks.