Large Graphs on multi-GPUs

Enrico Mastrostefano
Sapienza University of Rome
Sapienza University of Rome, 2013
@article{mastrostefano2013large,

   title={Large Graphs on multi-GPUs},

   author={Mastrostefano, Enrico},

   year={2013}

}

Download Download (PDF)   View View   Source Source   
We studied the problem of developing an efficient BFS algorithm to explore large graphs having billions of nodes and edges. The size of the problem requires a parallel computing architecture. We proposed a new algorithm that performs a distributed BFS and the corresponding implementation on multiGPUs clusters. As far as we know, this is the first attempt to implement a distributed graph algorithm on that platform. Our study shows how most straightforward BFS implementations present significant computation and communication overheads. The main reason is that, at each iteration, the number of processed edges is greater than the number actually needed to determine the parent or the distance array (the standard output of the BFS): there is always redundant information at each step. Reducing as much as possible this redundancy is essential in order to improve performances by minimizing the communication overhead. To this purpose, our algorithm performs, at each BFS level, a pruning procedure on the set of nodes that will be visited (NLFS). This step reduces both the amount of work required to enqueue new vertices and the size of messages exchanged among different tasks. To implement this pruning procedure efficiently is not trivial: none of the earlier works on GPU tackled that problem directly. The main issue being how to employ a sufficient large number of threads and balance their workload, to fully exploit the GPU computing power. To that purpose, we developed a new mapping of data elements to CUDA threads that uses a binary search function at its core. This mapping permits to process the entire Next Level Frontier Set by mapping each element of the set to one CUDA thread (perfect load-balancing) so the available parallelism is exploited at its best. This mapping allows for an efficient filling of a global array that, for each BFS level, contains all the neighbors of the vertices in the queue as required by the pruning procedure (based on sort and unique operations) of the array. This mapping is a substantial contribution of our work: it is quite simple and general and can be used in different contexts. We wish to highlight that it is this operation (and not the sorting) that makes possible to exploit at its best the computing power of the GPU. To speed up the sort and unique operations we rely on very efficient implementations (like the radix sort) available in the CUDA Thrust library. We have shown that our algorithm has good scaling properties and, with 128 GPUs, it can traverse 3 billion edges per second (3 GTEPS for an input graph with 228 vertices). By comparing our results with those obtained on different architectures we have shown that our implementation is better or comparable to state-of-the-art implementations. Among the operations that are performed during the BFS, the pruning of the NLFS is the most expensive in terms of execution time. Moreover, the overall computational time is greater then the time spent in communications. Our experiments show that the ratio between the time spent in computation and the time spent in communication reduces by increasing the number of tasks. For instance, with 4 GPUs the ratio is 2:125 whereas by using 64 GPUs the value is 1:12. The result can be explained as follows. In order to process the largest possible graph, the memory of each GPU is fully used and thus the subgraph assigned to each processor has a maximum (fixed) size. When the graph size increases we use more GPUs and the number of messages exchanged among nodes increases accordingly. To maintain a good scalability using thousands GPUs we need to further improve the communication mechanism that is, in the present implementation, quite simple. To this purpose, many studies employed a 2D partitioning of the graph to reduce the number of processors involved in communication. Such partitioning could be, in principle, implemented in our code and it will be the subject of a future work.
VN:F [1.9.22_1171]
Rating: 4.5/5 (2 votes cast)
Large Graphs on multi-GPUs, 4.5 out of 5 based on 2 ratings

You must be logged in to post a comment.

* * *

* * *

* * *

Free GPU computing nodes at hgpu.org

Registered users can now run their OpenCL application at hgpu.org. We provide 1 minute of computer time per each run on two nodes with two AMD and one nVidia graphics processing units, correspondingly. There are no restrictions on the number of starts.

The platforms are

Node 1
  • GPU device 0: AMD/ATI Radeon HD 5870 2GB, 850MHz
  • GPU device 1: AMD/ATI Radeon HD 6970 2GB, 880MHz
  • CPU: AMD Phenom II X6 @ 2.8GHz 1055T
  • RAM: 12GB
  • OS: OpenSUSE 11.4
  • SDK: AMD APP SDK 2.8
Node 2
  • GPU device 0: AMD/ATI Radeon HD 7970 3GB, 1000MHz
  • GPU device 1: nVidia GeForce GTX 560 Ti 2GB, 822MHz
  • CPU: Intel Core i7-2600 @ 3.4GHz
  • RAM: 16GB
  • OS: OpenSUSE 12.2
  • SDK: nVidia CUDA Toolkit 5.0.35, AMD APP SDK 2.8

Completed OpenCL project should be uploaded via User dashboard (see instructions and example there), compilation and execution terminal output logs will be provided to the user.

The information send to hgpu.org will be treated according to our Privacy Policy

HGPU group © 2010-2014 hgpu.org

All rights belong to the respective authors

Contact us:

contact@hgpu.org