D2K-Bench: Can LLM Agents Turn Expert Designs into Efficient GPU Kernels?
cs.LG, cs.AI, cs.DC
Submitted: 2026-10-02
Updated: 2026-10-02
Code: https://github.com/QwenLM/D2K-Bench
Sources
- Improving Efficiency of GPU Kernel Optimization Agents using a Domain-Specific Language and Speed-of-Light Guidance
- FastKernels: Benchmarking GPU Kernel Generation in Production
- SOL-ExecBench: Speed-of-Light Benchmarking for Real-World GPU Kernels Against Hardware Limits
- Harness Engineering for LLM-Driven GPU Kernel Generation
- Towards Automated Kernel Generation in the Era of LLMs
- MultiKernelBench: A Multi-Platform Benchmark for Kernel Generation
- DataKernelBench: Can LLMs Optimize Database Queries on GPUs?
- Towards Robust Agentic CUDA Kernel Benchmarking, Verification, and Optimization
- KernelBench-Verified: Do LLM-Generated Kernels Actually Beat PyTorch?
- KernelBenchX: A Comprehensive Benchmark for Evaluating LLM-Generated GPU Kernels
- AgentKernelArena: Generalization-Aware Benchmarking of GPU Kernel Optimization Agents
- RealisticTritonBench: A Benchmark for Triton-Kernel Generation in Real-World AI Frameworks
- Are LLM-Generated GPU Kernels Production-Ready? A Trace-Driven Benchmark and Optimization Agent
- AutoKernel: Autonomous GPU Kernel Optimization via Iterative Agent-Driven Search
- TritonForge: Profiling-Guided Framework for Automated Triton Kernel Optimization
- Astra: A Multi-Agent System for GPU Kernel Performance Optimization
- Making LLMs Optimize Multi-Scenario CUDA Kernels Like Experts
- KernelSkill: A Multi-Agent Framework for GPU Kernel Optimization
- DRTriton: Large-Scale Synthetic Data Driven Reinforcement Learning for Triton Kernel Generation
- Xe-Forge: Multi-Stage LLM-Powered Kernel Optimization for Intel GPU
Related papers
- Polynomial-Augmented Neural Networks (PANNs) with Weak Orthogonality Constraints for Enhanced Function and PDE Approximation
- AIRL-S: Unifying Reinforcement Learning and Search-Based Test-Time Scaling via Adversarial Inverse Reinforcement Learning
- Transformers as Bayesian In-Context Experimenters: Smoothness-Adaptive Efficient ATE Estimation
- Convergence issues in Relational Concept Analysis based on AOC-posets
- Beliefs Beyond Posteriors: The Predictive Cost of ELBO Binding
- Understanding Diffusion Models via Ratio-Based Function Approximation with SignReLU Networks