14283

Accelerating a Movie Recommender System Using VirtualCL on a Heterogeneous GPU Cluster

Ashish Bhatnagar
School of Information and Communication Technology, KTH Royal Institute of Technology, Stockholm, Sweden
KTH Royal Institute of Technology, 2015

@article{bhatnagar2015accelerating,

   title={Accelerating a Movie Recommender System Using VirtualCL on a Heterogeneous GPU Cluster},

   author={Bhatnagar, Ashish},

   year={2015}

}

Download Download (PDF)   View View   Source Source   

1577

views

Present day market offers a large number of movies which overwhelm people with choices. In order to quickly navigate through all the possible movies and find the interesting ones, the user can take advantage of recommender systems for movies. This thesis studies a movie recommender system which uses image processing and computer vision algorithms. The amount of time taken to analyze movies using these computation intensive algorithms is in the order of years. However, exploiting parallel nature of these algorithms using GPUs (Graphics Processing Unit) can help reduce the time many-folds. The primary goal of the thesis is to build a heterogeneous GPU cluster and use it to accelerate the algorithms of the recommender system. The guidelines and steps to build a heterogeneous GPU cluster given in the thesis can be used by other organizations and researchers. Results indicate that the heterogeneous GPU cluster platform can accelerate algorithms of a movie recommender system up to 5 times. The secondary goal of this thesis is to investigate the benefits of using VirtualCL framework which enables remote access to the GPUs of the cluster. Remote access to the GPUs provides energy efficiency and ease of cluster management. Results show that VirtualCL framework provides remote GPU capability at the cost of degradation in performance. Therefore, VCL framework should be used just for application areas where performance can be traded off for physical portability and ease of management.
Rating: 0.5/5. From 1 vote.
Please wait...

* * *

* * *

HGPU group © 2010-2024 hgpu.org

All rights belong to the respective authors

Contact us: