Configuration

We implement our GPU algorithms in C++ and run the experiments on an NVIDIA GeForce RTX 4090 with 24GB of memory using CUDA and compile with nvcc version 12.2.14.

This project uses GNU Make 4.3 as the build system. Make sure it is installed and available in your environment. You can check your version with:

make --version

GNU Make 4.3
Built for x86_64-pc-linux-gnu

To compile our code.

make

Then it will gereneate a execuable binary file named 'main'.

Execute

After compile successfully, replace {dataset} with the dataset name and {algorithm id} with algorithm id.

./main -d {dataset} -a {algorithm id} -t {warp start}

algorithm id:

• GPeel: 2
• Geld: 7
• obs1: 13
• obs2: 14
• obs3: 15
• Geld-Thread: 16
• Geld-Warp: 5
• Geld-Block: 17
• Geld-Block*: 18
• GHindex: 19

Example:

Run GPeel on EM dataset

./main -d em -a 2

Run Geld on EM dataset

./main -d em -a 7

Dataset

Please place the dataset inside the dataset folder. For example, if your dataset is named "em", create a subfolder within the dataset folder and store the em.txt file inside it.

Ensure that the name of the dataset folder matches the name of the .txt file.

e.g. dataset/em/em.txt

Dataset	Source
(AM)Amazon	https://snap.stanford.edu/data/index.html
(EM)Email-EuAll	https://snap.stanford.edu/data/index.html
(PO)Pokec	https://snap.stanford.edu/data/index.html
(SD)Slashdot	https://snap.stanford.edu/data/index.html
(EN)Enwiki-2024	https://law.di.unimi.it/index.php
(LJ)Live Journal	https://snap.stanford.edu/data/index.html
(UK)UK-2002	https://law.di.unimi.it/index.php
(H11)Hollywood-2011	https://law.di.unimi.it/index.php
(H09)Hollywood-2009	https://law.di.unimi.it/index.php
(IT)IT-2004	https://law.di.unimi.it/index.php
(IC)IndoChina-2004	https://law.di.unimi.it/index.php
(EU)EU-2015-TPD	https://law.di.unimi.it/index.php

profile

nsys profile --stats=true ./main -d {dataset} -a {algorithm id}

ncu --section "Occupancy" ./main -d {dataset} -a {algorithm id}