一句话：提出基于行为克隆的智能自动调优器 SparseZETA，单遍决策直接生成高性能 SpMV 程序，显著加速并大幅降低设计成本。
SpMV 在科学计算、图分析和机器学习中很关键，但性能高度依赖稀疏模式，常需定制程序。SparseZETA 通过模仿并组合人类专家操作来生成高性能 SpMV 程序：它将自动调优重构为行为克隆，根据部分程序设计的实时状态，以单遍决策顺序预测动作并直接合成程序，无需昂贵搜索；自训练机制加速训练数据收集。在 NVIDIA A100（及 RTX 2080 Ti）上，它相较现有自动调优器、人工程序和稀疏编译器取得平均 1.27×–15.66×（1.44×–19.07×）加速，并将稀疏格式创建与内核实现的设计时间从数天甚至数月降至每矩阵平均 82.52ms，仅需单 CPU 轻量推理。
Sparse matrix-vector multiplication (SpMV) is a crucial operation in scientific computing, graph analytics, and machine/deep learning. Its performance is highly sensitive to matrix sparsity patterns, necessitating tailored program designs. This paper introduces SparseZETA, an intelligent auto-tuner that generates high-performance, machine-designed SpMV programs by directly mimicking and composing human-expert actions. To efficiently navigate the vast design space, SparseZETA reformulates auto-tuning as a behavior-cloning problem: rather than costly exploration, it directly synthesizes programs by sequentially predicting actions in a one-pass decision-making process, guided by the real-time state of the evolving, partially constructed program designs. A novel self-training mechanism further accelerates the collection of training data for the prediction models. On NVIDIA A100 (and RTX 2080 Ti) GPUs, SparseZETA achieves average speedups of 1.27×–15.66× (1.44×–19.07×) over existing auto-tuners, human-designed programs, and a sparse compiler. SparseZETA substantially reduces the human effort required to design SpMV programs, including sparse format creation and kernel implementation, cutting the design time from days or even months to an average of 82.52ms per matrix via lightweight inference on only one CPU.
@article{DuLCZFCL26,
title = {SparseZETA: Intelligent Auto-tuner for Designing High-Performance SpMV Programs},
author = {Zhen Du and Ying Liu and Xionghui Chen and Yanbo Zhao and Xiaobing Feng and Huimin Cui and Jiajia Li},
journal = {Proceedings of the ACM on Programming Languages},
volume = {10},
number = {PLDI},
year = {2026},
doi = {10.1145/3808313}
}