一句话：提出 Hector 两级 IR 与代码生成框架，通过算子调度、数据访问和物化优化，显著加速 GPU 上的关系图神经网络。
关系图神经网络（RGNN）能建模异构图中的多类节点和边，但其内存密集型计算、编程接口与内核 API 之间的鸿沟，以及数据布局与异构性耦合带来的优化负担，造成性能和编程效率问题。本文提出 Hector，一个两级中间表示及代码生成框架：它捕获 RGNN 模型的关键属性，在算子间调度与物化中寻找减少访存的机会；生成灵活数据访问方案的代码以消除冗余拷贝；并将模型语义、数据布局和算子优化解耦。Hector 基于通用矩阵乘和节点/边遍历模板实现。在 DGL 与 OGB 异构图上的 RGCN、RGAT、HGT 模型中，Hector 相比先进公开系统最高取得 9.9 倍推理加速和 43.7 倍训练加速，且未触发 OOM；线性算子重排和紧凑物化进一步提升最高 3.8 倍。表达三个模型仅需 51 行代码，生成约 8K 行 CUDA/C++ 代码。
Relational graph neural networks (RGNNs) are graph neural networks with dedicated structures for modeling the different types of nodes and edges in heterogeneous graphs. While RGNNs have been increasingly adopted in many real-world applications due to their versatility and accuracy, they pose performance and system design challenges: inherent memory-intensive computation patterns, the gap between the programming interface and kernel APIs, and heavy programming effort required to optimize kernels caused by their coupling with data layout and heterogeneity. To systematically address these challenges, we propose Hector, a novel two-level intermediate representation and its code generator framework that (a) captures the key properties of RGNN models, and opportunities to reduce memory accesses in inter-operator scheduling and materialization, (b) generates code with flexible data access schemes to eliminate redundant data copies, and (c) decouples model semantics, data layout, and operators-specific optimizations from each other to reduce programming effort. By building on one general matrix multiply (GEMM) template and a node/edge traversal template, Hector achieves up to 9.9× speed-up in inference and 43.7× speed-up in training compared with the state-of-the-art public systems on select models, RGCN, RGAT and HGT, when running heterogeneous graphs provided by Deep Graph Library (DGL) and Open Graph Benchmark (OGB). In addition, Hector does not trigger any out-of-memory (OOM) exception in these tests. We also propose linear operator reordering and compact materialization to further accelerate the system by up to 3.8×. As an indicator of the reduction of programming effort, Hector takes in 51 lines of code expressing the three models and generates a total of 8K lines of CUDA and C++ code. Through profiling, we found that higher memory efficiency allows Hector to accommodate larger input and therefore attain higher throughput in forward propagation, while backward propagation is bound by latency introduced by atomic updates and outer products.
@inproceedings{WuH0HZNH24,
title = {Hector: An Efficient Programming and Compilation Framework for Implementing Relational Graph Neural Networks in GPU Architectures},
author = {Kun Wu and Mert Hidayetoğlu and Xiang Song and Sitao Huang and Da Zheng and Israt Nisa and Wen-Mei Hwu},
booktitle = {ASPLOS},
year = {2024},
doi = {10.1145/3620666.3651322}
}