Using Quiver for PyG Examples

Quiver is a GPU-optimized distributed library for PyG. It can speed up graph sampling and feature aggregation through GPU when running PyG examples.

Installation

Assuming you have installed PyTorch and PyG, you can install Quiver as follows:

pip install torch-quiver>=0.1.1

Usage

The API and design documentation of Quiver can be found here.

Examples

We provide several examples to showcase the usage of Quiver within PyG:

Single-GPU Training

The single-GPU example leverages Quiver's ability of (i) GPU-based graph sampling and feature aggregation, and (ii) GNN data caching algorithm (which cache hot data in GPU memory) while enabling fast access to CPU data using a Quiver shared tensor implementation:

python single_gpu_quiver.py

Multi-GPU Training

The multi-GPU example further leverages Quiver's ability of (i) distributing sampling and feature aggregation to multiple GPUs, and (ii) using multi-GPU memories to cache and replicate hot GNN data:

python multi_gpu_quiver.py

Distributed Training

A Quiver-based distributed PyG example is coming soon.