Abstract
The Travelling Salesman Problem (TSP) is one of the most studied combinatorial optimization problem which is significant in many practical applications in transportation problems. The TSP problem is NP-hard problem and requires large computation power to be solved by the exact algorithms. In the past few years, fast development of general-purpose Graphics Processing Units (GPUs) has brought huge improvement in decreasing the applications’ execution time. In this paper, we implement 2-opt and 3-opt local search operators for solving the TSP on the GPU using CUDA. The novelty presented in this paper is a new parallel iterated local search approach with 2-opt and 3-opt operators for symmetric TSP, optimized for the execution on GPUs. With our implementation large TSP problems (up to 85,900 cities) can be solved using the GPU. We will show that our GPU implementation can be up to 20x faster without losing quality for all TSPlib problems as well as for our CRO TSP problem.
Highlights
The travelling salesman problem (TSP) is a wellknown combinatorial optimization problem which is important in many practical applications from various fields such as transport, electronics and other engineering fields
The speed-up is calculated by the formula: tCPU /tGPU, if the speed-up is greater than 1 the Graphics Processing Units (GPUs) execution time is smaller than the central processing unit (CPU) execution time
The main reason to start working on the GPU implementations of the 2-opt and 3-opt local search is to investigate the possibility to deploy local search operators on data-flow multiprocessor architecture such as GPU
Summary
The travelling salesman problem (TSP) is a wellknown combinatorial optimization problem which is important in many practical applications from various fields such as transport, electronics and other engineering fields. The GPUs can perform much more floating point operations per second (flops) and have larger bandwidth rate compared to the CPUs, but on the other hand, GPUs cannot perform multiple tasks in parallel They are capable of running a same task on the different data concurrently and because of that not all problems can fit the GPUs paradigm and exploit their superior performance. A new parallel iterated local search approach for solving TSP with parallel 2-opt and 3-opt operators optimized for the execution on GPUs is presented. The biggest achievement of this research is that we have developed the 2-opt and 3-opt local search, executed on the GPU that can solve large problems (e.g. the biggest TSPlib problem with 85,900 cities).
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
