A comparison of many-threaded differential evolution and genetic algorithms on CUDA

The recent time has seen the rise of consumer grade massively parallel environments. Powerful GPUs and multi-core processors became widely available and easy to use programming APIs such as nVidia CUDA, OpenCL, and DirectCompute simplify the development of applications that can utilize them. In this environment, the nature inspired meta-heuristics can be in suitable cases implemented in parallel without additional costs. Backed by the power of modern GPGPUs, the meta-heuristics can be deployed to solve practical real world problems. In this paper, we compare differential evolution and genetic algorithms implemented on CUDA when solving the independent tasks scheduling problem.