# Hawk: Harnessing Hardware-Aware Knowledge for High-Performance NPU Kernel Generation

arXiv: https://arxiv.org/abs/2607.01590 · HTML: https://mlsanity.com/paper/2607.01590/

## Key facts

- Cited by 1 papers in the map (data through 14 Oct 2026).
- Builds on 26 of its 39 references that are in the map.
- First posted on arXiv 2 Jul 2026.

## Builds on (most cited first, 26 of 26)

- [GPT-4 Technical Report](https://mlsanity.com/paper/2303.08774/) (2023)
- [Qwen3 Embedding: Advancing Text Embedding and Reranking Through Foundation Models](https://mlsanity.com/paper/2506.05176/) (2025)
- [DeepSeek-V4: Towards Highly Efficient Million-Token Context Intelligence](https://mlsanity.com/paper/2606.19348/) (2026)
- [GLM-5: from Vibe Coding to Agentic Engineering](https://mlsanity.com/paper/2602.15763/) (2026)
- [Qwen3.5-Omni Technical Report](https://mlsanity.com/paper/2604.15804/) (2026)
- [A Theory of Emergent In-Context Learning as Implicit Structure Induction](https://mlsanity.com/paper/2303.07971/) (2023)
- [Kevin: Multi-Turn RL for Generating CUDA Kernels](https://mlsanity.com/paper/2507.11948/) (2025)
- [CUDA-L1: Improving CUDA Optimization via Contrastive Reinforcement Learning](https://mlsanity.com/paper/2507.14111/) (2025)
- [CUDA Agent: Large-Scale Agentic RL for High-Performance CUDA Kernel Generation](https://mlsanity.com/paper/2602.24286/) (2026)
- [CudaForge: An Agent Framework with Hardware Feedback for CUDA Kernel Optimization](https://mlsanity.com/paper/2511.01884/) (2025)
- [RAG-Fusion: a New Take on Retrieval-Augmented Generation](https://mlsanity.com/paper/2402.03367/) (2024)
- [MultiKernelBench: A Multi-Platform Benchmark for Kernel Generation](https://mlsanity.com/paper/2507.17773/) (2025)
- [STARK: Strategic Team of Agents for Refining Kernels](https://mlsanity.com/paper/2510.16996/) (2025)
- [TritonRL: Training LLMs to Think and Code Triton Without Cheating](https://mlsanity.com/paper/2510.17891/) (2025)
- [AVO: Agentic Variation Operators for Autonomous Evolutionary Search](https://mlsanity.com/paper/2603.24517/) (2026)
- [Injecting the BM25 Score as Text Improves BERT-Based Re-rankers](https://mlsanity.com/paper/2301.09728/) (2023)
- [AscendKernelGen: A Systematic Study of LLM-Based Kernel Generation for Neural Processing Units](https://mlsanity.com/paper/2601.07160/) (2026)
- [KernelBlaster: Continual Cross-Task CUDA Optimization via Memory-Augmented In-Context Reinforcement Learning](https://mlsanity.com/paper/2602.14293/) (2026)
- [Morest: Model-based RESTful API Testing with Execution Feedback](https://mlsanity.com/paper/2204.12148/) (2022)
- [MMMORRF: Multimodal Multilingual Modularized Reciprocal Rank Fusion](https://mlsanity.com/paper/2503.20698/) (2025)
- [AscendCraft: Automatic Ascend NPU Kernel Generation via DSL-Guided Transcompilation](https://mlsanity.com/paper/2601.22760/) (2026)
- [Does Few-Shot Learning Help LLM Performance in Code Synthesis?](https://mlsanity.com/paper/2412.02906/) (2024)
- [Towards Better Answers: Automated Stack Overflow Post Updating](https://mlsanity.com/paper/2408.09095/) (2024)
- [SkyByte: Architecting an Efficient Memory-Semantic CXL-based SSD with OS and Hardware Co-design](https://mlsanity.com/paper/2501.10682/) (2025)
- [Towards Cold-Start Drafting and Continual Refining: A Value-Driven Memory Approach with Application to NPU Kernel Synthesis](https://mlsanity.com/paper/2603.10846/) (2026)
- [The Incredible Shrinking Context... in a Decompiler Near You](https://mlsanity.com/paper/2409.11157/) (2024)

## Built on by (most cited first, 1 of 1)

- [From Experience to Expertise: Adoption-Aware Memory Learning for Data-Scarce NPU Kernel Synthesis](https://mlsanity.com/paper/2609.35568/) (2026)

## Abstract (arXiv)

Developing high-performance kernels for Neural Processing Units (NPUs) is a critical industry bottleneck, requiring developers to manually navigate implicit hardware constraints and strict memory hierarchies. While large language models offer immense automation potential, they fail catastrophically on NPUs due to a fundamental lack of hardware-specific priors. Naively transplanting code snippets from similar NPU kernels may pass the compiler, but it consistently triggers runtime crashes and performance degradation by blindly violating underlying hardware constraints. To overcome this, we introduce Hawk, a training-free framework that harnesses hardware-aware knowledge through three core modules: (1) Run-Time Knowledge Synthesis Module, which employs a Triple-Part Executable Knowledge Representation to inherently couple the error context with executable semantics; (2) Bottleneck-Aware Knowledge Retrieval Module, which implements a 2D-Retrieval paradigm to project queries into orthogonal syntactic and hardware-aligned semantic spaces; and (3) Effect-Driven Knowledge Distillation Module, which leverages LLM-driven semantic arbitration to continuously distill the knowledge by pruning errors and consolidating redundancies based on the empirical execution feedback. Extensive evaluations on real-world NPU workloads demonstrate that Hawk elevates generation accuracy from 49.4% to 80.0%, while achieving up to a 2.2x execution speedup over state-of-the-art baselines.