Solving knapsack problems on GPU
CNRS; LAAS; 7 avenue du Colonel Roche, F-31077 Toulouse, France
Computers & Operations Research (April 2011)
@article{boyer2011solving,
title={Solving knapsack problems on gpu},
author={Boyer, V. and El Baz, D. and Elkihel, M.},
journal={Computers & Operations Research},
issn={0305-0548},
year={2011},
publisher={Elsevier}
}
A parallel implementation via CUDA of the dynamic programming method for the knapsack problem on NVIDIA GPU is presented. A GTX 260 card with 192 cores (1.4GHz) is used for computational tests and processing times obtained with the parallel code are compared to the sequential one on a CPU with an Intel Xeon 3.0GHz. The results show a speedup factor of 26 for large size problems. Furthermore, in order to limit the communication between the CPU and the GPU, a compression technique is presented which decreases significantly the memory occupancy.
April 21, 2011 by hgpu
