A Parallel Immune Algorithm Based on Fine-Grained Model with GPU-Acceleration

Abstract

Fine-grained parallel immune algorithm (FGIA), though a popular and robust strategy for solving complicated optimization problems, is sometimes inconvenient to use as its population size is restricted by heavy data communication and the parallel computers are relatively difficult to use, manage, maintain and may not be accessible to most researchers. In this paper, we propose a FGIA method based on GPUacceleration, which maps parallel IA algorithm to GPU through the CUDA. We implement the IA on the base of the framework of genetic algorithm (GA), the analytical results demonstrate that the proposed method increases the population size, speeds up its execution and provides ordinary users with a feasible FGIA solution.

GPU computing in discrete optimization. Part II: Survey focused on routing problems

“…Thus we have the one-ant-per-thread and the one-ant-per-block schemes. Many papers implement either the former (Bai et al [5], You [96], Diego et al [27]) or the latter (Li et al [51], Uchida et al [89]). Only a few publications (Cecilia et al [14], Delévacq et al [25]) compare the two.…”

“…In the routing related GPU literature the apparent parallelism has lead to the two parallelization schemes of assigning one individual to one thread (coarse grained parallelism) (Chen et al [17]), and one individual to one block (fine grained parallelism) (Li et al [53], Fujimoto and Tsutsui [32]), see also Table 3. In some papers, different parallelization schemes are used for different operators.…”

“…Year Problem Algorithm Operators Li et al [53] 2009 TSP IEA PMX, mutation Chen et al [17] 2011 TSP GA crossover, 2-opt mutation Fujimoto and Tsutsui [32] 2011 TSP GA OX, 2-opt local search gene string move, Zhao et al [98] 2011 TSP IEA multi bit exchange Author Selection Immune next Population Li et al [53] Better Tournament Chen et al [17] Best Fujimoto and Tsutsui [32] Best Zhao et al [98] Best position Tournament Author GPU(s) Max. Speedup CPU code Li et al [53] GeForce 9600 GT 11.5 ?…”

“…Year Problem Algorithm GPU(s) Catala et al [13] 2007 OP ACO GeForce 6600 GT Bai et al [5] 2009 TSP multi colony MMAS GeForce 8800 GTX Li et al [51] 2009 TSP MMAS GeForce 8600 GT Wang et al [90] 2009 TSP MMAS Quadro Fx 4500 You [96] 2009 TSP ACO Tesla C1060 Cecilia et al [14] 2011 TSP ACO Tesla C2050 Delévacq et al [25] With the arrival of CUDA and OpenCL, programming the GPU became easier and consequently more papers studied ACO implementations on the GPU. In CUDA and OpenCL there is the basic concept of having a thread/workitem as basic computational element.…”

Solving very large instances of the scheduling of independent tasks problem on the GPU

“…Vidal and Alba also proposed a parallel version of CGA in a single GPU [48] and multiple ones [47]. Li et al designed a fine-grained parallel immune algorithm [24]. Finally, Pinel et al presented in [38] a parallel CGA for GPU architectures with a new recombination operator that is especially designed for good performance on the GPU.…”