KernelArc: A Multi-Agent Framework for GPU Kernel Optimization

KernelArc, a multi-agent framework for autonomous GPU kernel optimization, achieved first-place rankings on representative L1, L2, Quantization, and FlashInfer tasks at the public SOL-ExecBench leaderboard snapshot recorded on July 30, 2026, according to a new arXiv paper (arXiv:2608.17071v1). The framework, evaluated on NVIDIA H100 and B200 GPUs, coordinates strategy-specialized agents through conclusions-only shared memory, a deterministic benchmark guard, and plateau-triggered drafting, producing implementations that include custom BF16 GEMM, static cuBLASLt Expert-API configuration tables, fused mixture-of-experts backward, shape-gated decoder-layer fusion, native NVFP4 grouped-query attention, and paged prefill attention. The paper argues that shared multi-agent search broadens exploration and reaches stronger incumbents within a fixed candidate budget, though the value of individual coordination features depends on the kernel and optimization stage.

arXiv:2608.17071v1 Announce Type: new Abstract: We present KernelArc, a multi-agent framework for autonomous GPU kernel optimization across heterogeneous workloads. Strategy-specialized agents run in parallel and coordinate through conclusions-only shared memory, a deterministic benchmark guard, and read-only cross-agent state with plateau-triggered drafting. We evaluate \kernelarc{} on NVIDIA H100 and B200 GPUs using category-representative SOL-ExecBench workloads. The resulting implementations span custom BF16 GEMM, static cuBLASLt Expert-API configuration tables, fused mixture-of-experts backward, shape-gated decoder-layer fusion, native NVFP4 grouped-query attention, and paged prefill attention. At the public SOL-ExecBench leaderboard snapshot recorded on July~30, 2026, these submissions ranked first on representative L1, L2, Quantization, and FlashInfer tasks. The trajectories support the paper's central motivation: shared multi-agent search can broaden exploration and reach stronger incumbents within a fixed candidate budget, while the value of individual coordination features depends on the kernel and optimization stage.