arXiv:2606. 26453v1 Announce Type: new Abstract: We present KernelPro, a closed-loop multi-agent system that automatically generates, profiles, and iteratively optimizes GPU kernel code by integrating large language model (LLM) code generation with hardware profiler feedback and pluggable bottleneck detection tools.
arXiv:2607. 24762v1 Announce Type: new Abstract: Machine learning models are increasingly embedded in everyday software, and most of their runtime is spent in a small set of compute kernels such as matrix multiplication, convolution, and normalization.
arXiv:2604. 01489v2 Announce Type: replace Abstract: High-performance GPU kernels are critical to modern machine learning systems, yet developing them remains a manual, expert-driven process.
KernelFoundry is a hardware‑aware evolutionary framework that optimizes GPU kernels by combining MAP‑Elites quality‑diversity search, meta‑prompt evolution, and template‑based parameter tuning. It generates SYCL and CUDA kernels, outperforming baseline methods with an average 2.3× speedup on KernelBench. The system is distributed, supports remote hardware access, and offers a flexible user interface for diverse real‑world kernel generation tasks.
arXiv:2607. 14541v1 Announce Type: new Abstract: Existing GPU kernel generation benchmarks draw problems from synthetic or curated sources that diverge from deployed workloads.
KernelOPT is a multi‑agent system that optimizes GPU kernels generated by compilers like PyTorch Inductor by treating compiled models as structured artifacts. It preserves vendor library calls and focuses on Triton sub‑kernels, using five profiling‑guided LLM agents and a four‑gate verification cascade to ensure correctness and performance before re‑stitching the model. On 250 KernelBench problems, KernelOPT achieves geometric mean speedups of 1.40×, 1.15×, and 1.07× over torch.compile at three optimization levels.
KernelGenBench is a unified benchmark that evaluates large language models and agentic systems for generating efficient Triton kernels across diverse operator sources and hardware platforms. It covers 210 operators from PyTorch ATen, vLLM, and cuBLAS, and tests a 110‑operator subset on six different chips, consuming over 15 billion tokens in evaluation. The study finds that no single method dominates across all sources and platforms, with significant variations in correctness and performance depending on the operator source and hardware, and that agentic approaches require millions of tokens per successful operator.
arXiv:2604.18616v2 Announce Type: replace-cross Abstract: LLM coding agents can generate correct GPU kernels, but their performance still trails expert libraries. Reaching peak throughput requires co...
arXiv:2606. 02963v1 Announce Type: new Abstract: Production inference increasingly targets a heterogeneous mix of accelerators.
arXiv:2607. 16241v1 Announce Type: cross Abstract: Recent large language models (LLMs) can generate custom CUDA kernels that appear to outperform PyTorch on benchmarks such as KernelBench.
KernelOPT is a multi-agent system that optimizes GPU kernels generated by compilers like PyTorch Inductor by treating compiled models as structured artifacts. It preserves vendor library calls and focuses on Triton sub-kernels, using five profiling-guided LLM agents and a four-gate verification cascade to filter and validate candidates. On 250 KernelBench problems, KernelOPT achieves geometric mean speedups of 1.40×, 1.15×, and 1.07× over torch.compile at different optimization levels.
arXiv:2607. 27231v1 Announce Type: cross Abstract: Large language models (LLMs) have significantly increased the demand for efficient accelerator kernels, but kernel development remains a highly specialized and labor-intensive task.
AsmEvo is an agentic assembly-level optimizer that targets compiled AMDGPU code objects, reconstructing a reassemblable representation and applying low-level edits guided by a long-horizon agent. It rebuilds ABI-preserving optimized objects and verifies functional equivalence through differential testing against the original binary. Experiments show significant speedups—up to 1.35× geometric mean on MI308X and 1.18× on MI300X—while maintaining correctness.