Kernel generation for hardware accelerators such as GPUs and NPUs has become a proving ground for large language models (LLMs), and state-of-the-art systems raise correctness through pipelines that couple LLMs with agentic reinforcement learning and evolutionary search. Such pipelines generate, compile, and execute large numbers of candidate kernels, discarding most of them and forgoing the opportunity to distill failures into reusable knowledge. Many discarded candidates are near-miss operators that compile and run but fail numerical validation; each embodies genuine domain knowledge and a nontrivial investment in LLM inference, cross-compilation, and hardware execution. We argue for a paradigm shift: rather than regenerate, debug. Debugging is far more constrained than generating from scratch: the search space is small and feedback is dense. We present a domain-specific debug agent that addresses three core challenges in autonomous repair: mitigating knowledge scarcity through retrieved patterns and diagnostic instrumentation, ensuring integrity through anti-cheat detection and full-coverage evaluation, and controlling cost via convergence guards and bounded iteration. Debugging serves two complementary roles: it extends the capability frontier by recovering operators that repeated regeneration fails to produce, and it lowers cost per deliverable operator. Debug Pass@1 achieves 66.7% versus Regenerate Avg Pass@1's 25.9% and Regenerate Pass@3's 40.7%, while consuming 92.8% fewer tokens per success than three-trial regeneration. Component ablations show that the knowledge base drives recovery, while integrity gates reject 12.5-33.3% of the successes the workflow itself accepted.
Large language models (LLMs) have significantly increased the demand for efficient accelerator kernels, but kernel development remains a highly specialized and labor-intensive task. The recent rise of LLMs and agentic frameworks offers a promising pathway toward automatic kernel generation. However, despite rapid progress, there is still no comprehensive benchmark to rigorously evaluate LLM-generated kernels across diverse operator sources or heterogeneous hardware platforms. We present KernelGenBench, a unified benchmark for systematically evaluating LLM- and agent-generated Triton kernels across diverse operator sources and heterogeneous hardware platforms. It comprises two complementary sub-benchmarks: KernelGenBench-MS (Multi-Source), evaluating 210 operators from three sources beyond standard PyTorch-centric tasks, and KernelGenBench-MC (Multi-Chip), measuring performance portability across six heterogeneous hardware platforms using a 110-operator subset. Our large-scale evaluation, consuming over 15 billion tokens, shows: (1) agent-based methods consistently outperform pure LLM sampling methods, while cuBLAS operators are the most challenging across all methods; (2) generation performance varies significantly across hardware platforms, with even recent kernel-specialized agents experiencing severe cross-platform degradation (e.g., AutoKernel drops from 87% on NVIDIA to 25% on Platform E); (3) autonomous kernel generation remains highly cost-intensive, with specialized agent methods averaging 5.11 million tokens per successful operator (AKO4all reaches 5.19 million), orders of magnitude higher than simple LLM sampling approaches.
Existing GPU kernel generation benchmarks draw problems from synthetic or curated sources that diverge from deployed workloads. We present Atrex-Bench, a benchmark whose 30 operators and 440 shapes are sampled directly from full-cluster production inference traces of compute-limited, memory-rich GPUs. Each problem carries an importance weight derived from its share of observed GPU time, weighted by application card-hours and computed separately for the serving phases in which it runs, together with a per-problem roofline ceiling, so the aggregate score emphasizes the kernels that consume the most serving time. Evaluating six frontier coding agents on Atrex-Bench shows that even the best vanilla model reaches only ${\sim}10\%$ of the hardware roofline on production operators; and correctness alone overstates capability, since much of the apparent pass rate comes from PyTorch fallbacks rather than kernels the model wrote. To close this gap, we co-release Atrex-Kernel-Agent (AKA), a profile-driven kernel-optimization agent that combines iterative measure-revise search, optimization dropout for escaping stalled search contexts, and a layered GPU-optimization knowledge base (298 reference-kernel files and 244 optimization-knowledge documents, plus external upstream reference projects for API/ISA lookup). In a controlled case study, the agent converts zero-FlyDSL fallbacks into real kernels that match or exceed hand-tuned production baselines.
Recent agentic approaches to LLM-based kernel generation have achieved impressive results on CUDA. For emerging AI accelerators such as AWS Trainium and Inferentia, automated kernel generation and optimization remain largely unaddressed. Writing kernels for these chips via the Neuron Kernel Interface (NKI) is particularly challenging: developers must navigate a multi-engine architecture, tile-based programming, and explicit data movement across multi-level memory hierarchy. Moreover, no publicly-available training data, benchmarks, or tool-augmented agents exist for this domain. We introduce NKI-Agent, the first system combining domain-specific supervised fine-tuning (SFT) with a compile-verify-fix agent loop for NKI kernel generation. We adapt the existing CUDA-Agent framework to Neuron hardware, curate 6,000 NKI kernel generation tasks for training, and construct NKIBench, a 250-task benchmark across three difficulty levels. Evaluated on real Trn1 hardware, NKI-Agent with Claude Opus 4.8 and a rank-aware system prompt achieves a 77.3% pass rate on the 150-task NKIBench. We show that tool use is critical: Opus 4.8 scores 6% in single-shot mode without agent tools. On a 60-task subset, we show that an SFT-trained Qwen3-Coder-30B-A3B achieves 25.0% pass rate at 1/100th the cost, outperforming Claude Sonnet 4 (15.0%). We also report that Group Relative Policy Optimization (GRPO) with binary compilation reward fails to improve over SFT, providing guidance on reward design for RL-based kernel generation.
High-performance GPU kernels are critical for reducing the exponentially growing computational costs of large language models (LLMs), but their development heavily relies on manual tuning by domain experts. While recent advances in LLM-based approaches show promise for automating kernel generation, they still struggle to achieve both correctness and high performance. This limitation primarily arises from the lack of domain-specific optimization guidance, hindering effective exploration of the optimization space. We propose EGG, an Expert-Guided Agent Framework for Kernel Generation, which incorporates expert optimization principles to guide LLMs' decisions. Inspired by expert workflows, we decompose kernel generation into two hierarchical stages: 1) algorithmic structure design, which establishes a high-quality computational structure foundation; 2) hardware-specific tuning, which performs targeted adjustments through parallel mapping, tensor tiling, and memory optimization. This staged decomposition defines explicit optimization objectives, structuring the design space to achieve progressive refinement. To this end, a stage-aware multi-agent collaboration mechanism is designed for inter and intra-stage context management, ensuring stable optimization trajectories. Experiments on KernelBench and real-world workloads show that EGG achieves a 2.13x average speedup over PyTorch, outperforming existing agent-based and RL-based approaches.
Jason Yoo, Rajarshi Saha, Shaowei Zhu +3cs.AI cs.MA
Despite rapid progress in LLM-based code generation, writing correct and performant kernels for hardware accelerators remains a key bottleneck in scaling modern ML workloads. We present MKEvolve (Modular Kernel Evolve), a framework that iteratively co-evolves a modular decomposition of complex PyTorch modules and the LLM-generated kernel for each submodule, refining the decomposition by splitting and fusing across iterations while independently improving each subkernel via LLM-driven beam search. The resulting kernels are programmatic compositions of independently verified subkernels, making them configurable (subkernel implementations are swappable), interpretable (errors and speedups are traceable to specific subkernels), and readily adaptable to related model architectures. Experiments with Triton on KernelBench L2 and L3, spanning multi-operator sequences and full model architectures, show that MKEvolve improves both correctness and speedup over end-to-end direct synthesis baselines while reducing LLM token usage by up to 35%.