High-density package integration has become the trend of modern high-end packaging, which achieve faster signal transmission speed through heterogeneous integration or homogeneous integration, such products are usually used in server, high performance computing or net-working switch, etc. In addition to the difficulty of the process, the accompanying challenges also have more severe thermal problems. Take server for example, the thermal design power (TDP) of the server is increasing year by year, and according to the information on the internet, it can basically reach more than 500W per package. If there is no effective cooling methold, the reliability of the packag e itself will be affected under such severe operating conditions. It can usually reduce the temperature of the IC with a powerful cooling system, such as general water cooling system, single-phase or two-phase immersion water cooling system. Under these such powerful cooling system, the junction temperature of IC is no longer the problem. In this study, computational fluid dynamics (CFD) models of the general water cooling system is constructed and calibrated using the normal FCBGA PKG firstly. The simulation results are shown the good alignment that error is below 6%. A CFD model of a Fan-Out Embedded Bridge (FOEB) package is then established to investigate the thermal performance under general water cooling system with different PKG design through parametric analysis. The results show that no matter how the thermal performance of PKG level is optimized by structure or material, such as compound type, top die thickness and the substrate design, there is no significant difference in the overall junction temperature performance. On the contrary, by changing the system level parameters, such as thermal interface material (TIM) that ouside the PKG, flow velocity that inside the channel, etc, it is more effective to see the obvious difference in IC temperature. Overall, this research proposes the optimize result for the general water cooling system through simulation, and compares the degree of influence at the package and system level.