A Comprehensive Survey on Various Evolutionary Algorithms on GPU
Satvir Singh, Jaspreet Kaur, Rashmi Sharan Sinha · 2014
This paper presents a comprehensive survey on parallelizing computations involved in optimization problem on Graphics Processing Unit (GPU) using CUDA (Compute Unified Design Architecture). GPU have multithread cores with high memory bandwidth which allow for greater ease of use and also more radially support a layer body of applications. Many researchers have reported significant speedups with General Purpose computing on GPU (GPGPU). Stochastic meta-heuristic search algorithms, e.g., Mixed Integer NonLinear Programming (MINLP), Central Force Optimization(CFO), Genetic Algorithms (GA), and Particle Swarm Optimization(PSO), etc. are being investigated nowadays for improved performance with processing power of GPU. From study it is found that GPGPU shows tremendous speedups from 7 times in Steady State GAs to 10, 000 times speedups in CFO.