In this study, the numerical simulation of convective heat transfer of nanofluids (Al2O3/water and CuO/water) inside a sinusoidal wavy channel is performed using the Graphics Processing Units (GPU). The governing equations including stream-function, vorticity transport, and energy are discretized using the fourth-order Spline Alternating-Direction Implicit (SADI) approach in combination with the curvilinear coordinates mapping. The final tridiagonal-matrices are solved by Parallel-Thomas-Algorithm (PTA) on GPU. A homogenous one-phase model is also applied to consider the effective characteristics of the nanofluids flow. In the first part of the results section, the effects of nanoparticle volume fraction, Reynolds number, and amplitude of the wavy wall on average Nusselt number and the skin friction coefficient of the channel are investigated. In the second part of the results section, the ability of GPU to accelerate the computation (or runtimes) is compared to the classic Thomas algorithm that runs on a Central Processing Unit (CPU). The results demonstrate that the speedup of PTA against CPU runtime for the finest grid is around 18.32×.