search

SJTU-IPADS / gnnlab

A Factored System for Sample-based GNN Training over GPUs
Apache License 2.0
42 stars 7 forks source link

readme

GNNLab

GNNLab is a factored system for sample-based GNN training over GPUs. GNNLab dedicates each GPU to the task of graph sampling or model training. It accelerates both tasks by eliminating GPU memory contention. Furthermore, GNNLab embodies a new pre-sampling based caching policy that takes both sampling algorithms and GNN datasets into account, showing an efficient and robust caching performance. This work is cooperated with the GraphScope team from Alibaba Group.

gs-badge-cowork

Paper

GNNLab: A Factored System For Sample-based GNN Training Over GPUs, EuroSys'22, Jianbang Yang, Dahai Tang, Xiaoniu Song, Lei Wang, Qiang Yin, Rong Chen, Wenyuan Yu, and Jingren Zhou.

Artifact Evaluation for GNNLab: Link.

Terminology

FGNN is the initial version GNNLab, while SGNN is the initial version of TSOTA, a baseline system. SamGraph is the framework shared by the above system.

Table of Contents

Project Structure

> tree .
├── datagen                     # Dataset Preprocessing
├── example
│   ├── dgl
│   │   ├── multi_gpu           # DGL models
│   ├── pyg
│   │   ├── multi_gpu           # PyG models
│   ├── samgraph
│   │   ├── balance_switcher    # FGNN Dynamic Switch
│   │   ├── multi_gpu           # FGNN models
│   │   ├── sgnn                # SGNN models
│   │   ├── sgnn_dgl            # DGL PinSAGE models(SGNN simulated)
├── exp                         # Experiment Scripts
│   ├── figXX
│   ├── tableXX
├── samgraph                    # FGNN, SGNN source codes
└── utility                     # Useful tools for dataset preprocessing

Paper's Hardware Configuration

Installation

Software Version

Install CUDA10.1

FGNN is built on CUDA 10.1. Follow the instructions in https://developer.nvidia.com/cuda-10.1-download-archive-base to install CUDA 10.1, and make sure that /usr/local/cuda is linked to /usr/local/cuda-10.1.

Install GCC-7

CUDA10.1 requires GCC (version<=7). Hence, make sure that gcc is linked to gcc-7, and g++ is linked to g++-7. 

```bash
# Ubuntu
sudo update-alternatives --install /usr/bin/gcc gcc /usr/bin/gcc-7 7
sudo update-alternatives --install /usr/bin/g++ g++ /usr/bin/g++-7 7
```

Install GNN Training Systems

We use conda to manage our python environment.

  1. We use conda to manage our python environment.

    conda create -n fgnn_env python==3.8 pytorch==1.7.1 torchvision==0.8.2 torchaudio==0.7.2 cudatoolkit=10.1 -c pytorch -y
conda activate fgnn_env
conda install cudnn numpy scipy networkx tqdm pandas ninja cmake -y # System cmake is too old to build DGL
sudo apt install gnuplot # Install gnuplot for experiments:

  2. Download GNN systems.

    # Download FGNN source code
git clone --recursive https://github.com/SJTU-IPADS/gnnlab.git

  3. Install DGL, PyG, and FastGraph. The package FastGraph is used to load datasets for GNN systems in all experiments.

    # Install DGL
./gnnlab/3rdparty/dgl_install.sh

# Install PyG
./gnnlab/3rdparty/pyg_install.sh

# Install fastgraph
./gnnlab/utility/fg_install.sh

  4. Install FGNN (also called SamGraph) and SGNN.

    cd gnnlab
./build.sh

Change ULIMIT

Both DGL and FGNN need to use a lot of system resources. DGL CPU sampling requires cro-processing communications while FGNN's global queue requires memlock(pin) memory to enable faster memcpy between host memory and GPU memory. Hence we have to set the user limit.

Append the following content to /etc/security/limits.conf and then reboot:

* soft nofile 65535         # for DGL CPU sampling
* hard nofile 65535         # for DGL CPU sampling
* soft memlock 200000000    # for FGNN global queue
* hard memlock 200000000    # for FGNN global queue

After reboot you can see:

> ulimit -n
65535

> ulimit -l
200000000

Docker Support

We provide two methods to get the Docker image.

  1. Build the image from Dockerfile

    We provide a Dockerfile to build the experiment image. The file is in the root directory of this repository. Users can use the following command to create a Docker environment.

    docker build -t gnnlab/fgnn:v1.0 .

  2. Pull the image from Docker Hub (about 4.24GB)

    docker pull gnnlab/fgnn:v1.0
# We only test this image in the platform with all V100 GPUs
# If users meet some running problems on other GPUs, we provide all tools needed in the image to re-build DGL, FGNN and PyG.

Then users can run tests in Docker.

docker run --ulimit memlock=-1 --rm --gpus all -v $HOST_VOLUMN:/graph-learning -it gnnlab/fgnn:v1.0 bash

Make sure that Docker can support CUDA while building images. Here is a reference to solve Docker building images with CUDA support.

Dataset Preprocessing

See datagen/README.md to find out how to preprocess datasets.

QuickStart: Use FGNN to train GNN models

FGNN is compiled into Python library. We have written several GNN models using FGNN’s APIs. These models are in gnnlab/example and are easy to run as following:

cd gnnlab/example

python samgraph/multi_gpu/train_gcn.py --dataset papers100M --num-train-worker 1 --num-sample-worker 1 --pipeline --cache-policy pre_sample --cache-percentage 0.1 --num-epoch 10 --batch-size 8000

Experiments

Our experiments have been automated by scripts (run.py). Each figure or table in our paper is treated as one experiment and is associated with a subdirectory in gnnlab/exp. The script will automatically run the experiment, save the logs into files, and parse the output data from the files.

Note that running all experiments may take several hours. This table lists the expected running time for each experiment.

See exp/README.md to find out how to conduct the experiments.

License

FGNN is released under the Apache License 2.0.

Academic and Conference Papers

[EuroSys] GNNLab: A Factored System for Sample-based GNN Training over GPUs. Jianbang Yang, Dahai Tang, Xiaoniu Song, Lei Wang, Qiang Yin, Rong Chen, Wenyuan Yu, Jingren Zhou. Proceedings of the 17th European Conference on Computer Systems, Rennes, France, April, 2022.