A new approach of the crankpin bearing (CB) surface designed by different partial textures (PTs) with spherical dimples (SDs), conical dimples (CDs), or circular-cylindrical dimples (CCDs) is proposed to enhance the lubrication performance and reduce the friction power loss of engines. To assess the effect of the geometric dimensions of different PTs, such as the distribution density, depth, and shape of dimples on the lubrication performance and friction power loss (LP-FPL), a hydrodynamic model of the engine’s CB lubrication is then built. Then a genetic algorithm (GA) developed in MATLAB software is applied for optimizing the geometric dimensions of the PTs to further improve the LP-FPL. Both the oil film pressure and friction force of the CB are chosen as the objective functions to evaluate the LP-FPL of engines. The obtained results show that the LP-FPL of engines with the CB surface designed by PTs at the mixed lubrication regime is significantly improved in comparison with the CB surface designed by full textures. In addition, the PTs designed by the SDs with a distribution density of n = 4 and m = 6 in the x and y directions of the bearing surface improve the LP-FPL better than other structures of conical dimples (n = 4 and m = 6) [CDs(4 × 6)], circular-cylindrical dimples (n = 4 and m = 6) [CCDs(4 × 6)], spherical dimples (n = 3 and m = 6) [SDs(3 × 6)], and spherical dimples (n = 5 and m = 6) [SDs(5 × 6)]. Particularly, at an engine speed of 2000 revolutions per minute (rpm), the maximum values of the oil film pressure and friction force with the spherical dimples (n = 4 and m = 6) [SDs(4 × 6)] optimized by the GA are remarkably improved by 7.2% and 22.7% in comparison with full textures. Therefore, the use of the optimal SDs(4 × 6) could further enhance the LP-FPL of engines.