# EGraph

**Repository Path**: GPGPM/EGraph

## Basic Information

- **Project Name**: EGraph
- **Description**: No description available
- **Primary Language**: C++
- **License**: Not specified
- **Default Branch**: master
- **Homepage**: None
- **GVP Project**: No

## Statistics

- **Stars**: 0
- **Forks**: 0
- **Created**: 2021-02-01
- **Last Updated**: 2021-02-22

## Categories & Tags

**Categories**: Uncategorized

**Tags**: None

## README

## EGraph

It is an efficient system to effectively support concurrent execution of TGP jobs on large-scale dynamic graphs with the help of GPU accelerators. Specifically, it proposes an effective Loading-Processing-Switching (LPS) execution model to effectively reduce the overhead of CPU-GPU data transfer by fully utilizing the data access similarity between the TGP jobs, thereby ensuring a higher GPU utilization ratio for efficient execution of the TGP jobs. Besides, it can be integrated into the existing GPU-accelerated graph processing systems without much effort.

#### Graph formats

EGraph accepts the binary serialized pre-built CSR graph representation. Reading binary formats is faster and more space efficient.

#### Compilation

To compile EGraph, just run make in the root directory. The only requrements are g++ and CUDA toolkit.

#### Running applications in EGraph

The applications take a graph as input as well as some optional arguments. For example:

```
$ ./sssp-TGP --input path-to-input-graph
$ ./sssp-TGP --input path-to-input-graph --source 10