This paper presents experimental and numerical studies aimed at evaluating and comparing the performances of heat sinks for liquid cooling of a heterogeneous heat-generating surface with multiple heat sources. Various heat sink prototypes were optimized, machined, instrumented, and tested, including a uniform straight channel (RSC) heat sink as the baseline case, an optimized straight channel (OSC) heat sink, and a genetic algorithm-based topology optimization (GATO) heat sink. Infrared (IR) thermography was employed to measure the near-wall fluid temperature field in the heat sink, facilitated by introduction and installation of a sapphire window. The detailed spatial temperature distribution obtained enabled the analysis of heat transfer characteristics at the local level, with the good agreement between CFD results and IR measurement providing a solid validation of the numerical simulation models. Following this experimental validation, a systematic numerical study was conducted to evaluate the thermal and hydraulic performances of the three heat sinks under a wide range of operating conditions. Results showed that the GATO heat sink consistently outperforms the RSC and OSC heat sinks, exhibiting superior global thermal performances. This was evidenced by its better temperature hotspot removal capability, higher Nu number, higher PEC number, and higher Le Goff number compared to the other heat sinks. The effectiveness and robustness of the GATO approach for heat sink optimization were thereby proven, highlighting its significant potential in addressing general thermal management issues.