Solving knapsack problems on GPU

Abstract

International audienceIn this article, we propose a parallel implementation of the dynamic programming method for the knapsack problem on NVIDIA GPU. A GTX 260 (192 cores, 1.4GHz) was 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 up to 26 and permit one to solve large size problems within a reasonable processing time. Furthermore, in order to limit the communication between the CPU and the GPU, a compression technique is presented which decreases significantly the memory occupancy

Dense Dynamic Programming on Multi GPU

“…We believe that further speedup can be obtained on multi GPU clusters with extensive use of texture memory in GPUs and by improving data compression techniques, just as in [8].…”

“…In [8], we have studied the parallel implementation via CUDA of the dense dynamic programming method on a CPU/GPU system with a 3 GHz Xeon Quadro Intel processor and a GTX 260 GPU card.…”

“…We refer to [8] for aspects related to memory occupancy and data compression techniques. Reference is also made to [9], for a study on local search methods and GPU computing for combinatorial optimization problems.…”

“…In [8], we have presented an original implementation via CUDA of the dense dynamic programming method for KP on a CPU/GPU system with a single GPU. Experiments carried out on a CPU with 3 GHz Xeon Quadro Intel processor and GTX 260 GPU card have shown substantial speedup.…”

GPU Implementation of the Branch and Bound Method for Knapsack Problems

“…These codes will be combined in order to provide efficient parallel hybrid methods. In [11], we have proposed an implementation of the dynamic programming method for KP on a CPU/GPU system via CUDA. Experiments carried out on a CPU with 3 GHz Xeon Quadro Intel processor and GTX 260 GPU have shown substantial speedup.…”