Rapid developments in the design of chips and electronic devices for high-performance computers have led to a need for new and more effective methods of chip cooling. The first purpose of this study was to investigate the thermal development and heat transfer characteristics of aluminum foam heat sinks for the heater which simulated Intel core i7 processor. Three main features were then added to the aluminum foam heat sink: 1) introducing pulsating water flow through the aluminum foam in order to achieve a uniform surface temperature, 2) the addition of channels in the aluminum foam in order to increase the surface area to volume ratio and 3) using γ-Al2O3-water nanofluid as a coolant instead of water. The experimental results revealed that the thermal entry length of the flow through the aluminum foam increases along with increases in the Reynolds number. The results also revealed a convex profile for the local temperature distribution of the pulsating water flow through the aluminum foam due to the reversing flow and development of a boundary layers. The pulsating flow also enhanced the average Nusselt number by 14% and the temperature uniformity by 73% compared to the steady flow. The introduction of channels in the aluminum foam reduced the average Nusselt number by 10% (for two channels) and 25% (for three channels). The results revealed that the aluminum foam with two channels achieved a higher thermal efficiency compared to the block and three channel designs. The results also revealed that the foam filled channel enhanced the average Nusselt number by 20% compared with the empty channel. The maximum heat transfer rate enhancement was achieved at 0.2vol% and there was a sudden drop in the positive effect at 0.3vol% (compared with pure water). The positive effect then showed a slight increase along with increases in nanoparticle concentration up to 0.6vol%. The average enhancement percentages of the Nusselt number at a 0.2vol% nanofluid concentration were 37% and 28% at Reynolds numbers of 601.3 and 210, respectively. The numerical results were in good agreement with the experimental data with a maximum relative error of 3% in all studies.