An efficient Immersed Boundary Method (IBM) for solving conjugate heat transfer (CHT) problems is proposed. The conventional approach of computing heat flux on the fluid-solid surface to derive temperature as the boundary condition is impractical for parallel computation. To overcome this limitation and optimize computational efficiency, a strongly coupled IBM strategy for CHT is developed based on our previous work [1,2], employing a hierarchical data structure to represent complex geometry in a supercomputer environment. The concept is that, instead of calculating heat flux on the fluid-solid interface, the energy conservation principle on the interface cells is utilized. By solving the continuity and energy equations in the Navier-Stokes equation on these cells, temperature can be determined under the condition of energy conservation. Furthermore, for the application of this IBM to modern thermal problems, the framework's capability to handle thermal phenomena with significant heat transfer is considered. Consequently, a fully compressible solver is adopted to compute density variation. The governing equation is unified for thermal conduction inside the solid and thermal convection in the fluid by modyfing velocity and pressure for the preconditioned Navier-Stokes equation. The resulting framework is highly efficient and suitable for fluid flow with significant heat transfer, making it applicable to addressing complex phenomena in practical thermal engineering problems.