By a rapid development of graphics processing unit (GPU), the programmability and highly parallel processing feature of GPU create a chance to allow the general purpose computation to be conducted on GPU, conventionally called GPGPU (general purpose computation on GPU). A brief survey, in particular on the rationale of how the GPU architecture leads to GPGPU, is introduced in this paper, and various applications by taking advantage of GPGPU computation, including our works in this field on fluid simulation, as well as the works to be presented in this special session are also introduced.