FeatGraph: Sparse kernels for GNNs based on TVM Graph neural networks (GNNs) are gaining popularity in recent years as a promising approach to machine learning on graphs. Unlike traditional graph workloads where each vertex/edge is associated with a scalar, GNNs attach a feature tensor to each vertex/edge. WebTo tackle the challenge, FeatGraph maps the building blocks of GNNs to generalized SpMM (sparse-dense matrix multiplication) and SDDMM (sampled dense-dense matrix multiplication) kernels, and provides high-performance implementations of these sparse kernels based on TVM. For more information, refer to our SC'20 paper.
FeatGraph: A Flexible and Efficient Backend Graph …
Webdgl / featgraph / src / featgraph.cc Go to file Go to file T; Go to line L; Copy path Copy permalink; This commit does not belong to any branch on this repository, and may belong to a fork outside of the repository. Cannot retrieve … WebAug 26, 2024 · FeatGraph incorporates optimizations for graph traversal into the sparse templates and allows users to specify optimizations for UDFs with a feature dimension … the state is a permanent object
FeatGraph: A Flexible and Efficient Backend for Graph Neural …
WebAug 25, 2024 · FeatGraph incorporates optimizations for graph traversal into the sparse templates and allows users to specify optimizations for UDFs with a feature dimension … WebNov 1, 2024 · FeatGraph: A Flexible and Efficient Backend for Graph Neural Network Systems DOI: 10.1109/SC41405.2024.00075 Authors: Yuwei Hu Ye Zihao Fudan University Minjie Wang Jiali Yu Show all 9 authors No... Web2 days ago · Boston has already become the fastest team to 50 wins and 100 points, and the team broke the record for most wins in a season on Sunday with its 63rd win. The B's also required the fourth-fewest ... the state is me quote