# DataKernelBench: Can LLMs Optimize Database Queries on GPUs?

arXiv: https://arxiv.org/abs/2608.25061 · HTML: https://mlsanity.com/paper/2608.25061/

## Key facts

- Cited by 1 papers in the map (data through 14 Oct 2026).
- Builds on 22 of its 43 references that are in the map.
- First posted on arXiv 25 Aug 2026.

## Builds on (most cited first, 22 of 22)

- [Constitutional AI: Harmlessness from AI Feedback](https://mlsanity.com/paper/2212.08073/) (2022)
- [LiveCodeBench: Holistic and Contamination Free Evaluation of Large Language Models for Code](https://mlsanity.com/paper/2403.07974/) (2024)
- [gpt-oss-120b & gpt-oss-20b Model Card](https://mlsanity.com/paper/2508.10925/) (2025)
- [DeepSeek-V4: Towards Highly Efficient Million-Token Context Intelligence](https://mlsanity.com/paper/2606.19348/) (2026)
- [GLM-4.5: Agentic, Reasoning, and Coding (ARC) Foundation Models](https://mlsanity.com/paper/2508.06471/) (2025)
- [Text-to-SQL Empowered by Large Language Models: A Benchmark Evaluation](https://mlsanity.com/paper/2308.15363/) (2023)
- [KernelBench: Can LLMs Write Efficient GPU Kernels?](https://mlsanity.com/paper/2502.10517/) (2025)
- [The Case for Learned Index Structures](https://mlsanity.com/paper/1712.01208/) (2017)
- [Neo: A Learned Query Optimizer](https://mlsanity.com/paper/1904.03711/) (2019)
- [Kevin: Multi-Turn RL for Generating CUDA Kernels](https://mlsanity.com/paper/2507.11948/) (2025)
- [TritonBench: Benchmarking Large Language Model Capabilities for Generating Triton Operators](https://mlsanity.com/paper/2502.14752/) (2025)
- [Evaluating Language Models for Efficient Code Generation](https://mlsanity.com/paper/2408.06450/) (2024)
- [Towards Robust Agentic CUDA Kernel Benchmarking, Verification, and Optimization](https://mlsanity.com/paper/2509.14279/) (2025)
- [CUDA Agent: Large-Scale Agentic RL for High-Performance CUDA Kernel Generation](https://mlsanity.com/paper/2602.24286/) (2026)
- [Devstral: Fine-tuning Language Models for Coding Agent Applications](https://mlsanity.com/paper/2509.25193/) (2025)
- [TritonRL: Training LLMs to Think and Code Triton Without Cheating](https://mlsanity.com/paper/2510.17891/) (2025)
- [Query Processing on Tensor Computation Runtimes](https://mlsanity.com/paper/2203.01877/) (2022)
- [Share the Tensor Tea: How Databases can Leverage the Machine Learning Ecosystem](https://mlsanity.com/paper/2209.04579/) (2022)
- [Bespoke OLAP: Synthesizing Workload-Specific One-size-fits-one Database Engines](https://mlsanity.com/paper/2603.02001/) (2026)
- [Rethinking Analytical Processing in the GPU Era](https://mlsanity.com/paper/2508.04701/) (2025)
- [GPU Acceleration of SQL Analytics on Compressed Data](https://mlsanity.com/paper/2506.10092/) (2025)
- [PystachIO: Efficient Distributed GPU Query Processing with PyTorch over Fast Networks & Fast Storage](https://mlsanity.com/paper/2512.02862/) (2025)

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

- [D2K-Bench: Can LLM Agents Turn Expert Designs into Efficient GPU Kernels?](https://mlsanity.com/paper/2610.03226/) (2026)

## Abstract (arXiv)

GPUs increasingly accelerate database systems, but query-specific peak performance still often relies on hand-written kernels. Existing LLM kernel benchmarks focus on machine learning operators, leaving irregular, heterogeneous, data-movement-heavy database-style operators untested. We introduce DataKernelBench, which translates SQL into validated PyTorch TorchPlan programs and evaluates LLMs that optimize either the core tensor-bounded snippet or the full query in CUDA or Triton through execution-guided repair. Across ten proprietary and open-weight models on TPC-H SF10 with an H100 GPU, the strongest full-query CUDA configuration achieves 2.11× speedup over the TorchPlan baseline at full pass rate. We find that higher-performing implementations commonly use kernel fusion and execution-strategy changes, stronger models benefit most from full-query specialization, and workload context matters more than hardware context. To handle data larger than GPU memory, we extend TorchPlan with Dask-cuDF for on-demand partition loading on TPC-H SF100 with four H100 GPUs, achieving 2.54× speedup. Project page: https://kerneldf.github.io/datakernelbench