Abstract:
The present paper illustrates experimentally the impact of non-uniform heat generation in a microprocessor on
the hot spot distribution and the method to cool the same, efficiently employing parallel microchannel cooling
configurations. It is often assumed during design of microprocessor cooling systems that the heat load emitted by
the device is uniform, however real time tracking of the device shows that such assumptions are far from the
reality. An Intel® Core™ i7–4770 3.40 GHz quad core processor has been experimentally mimicked using heat
load data retrieved from a real microprocessor with non-uniform core activity. Parallel microchannel based heat
spreader configurations using U, I and Z type flow configurations have been employed to mitigate the thermal
load from the mimicked device. The observations clearly show that the microchannel cooling system experiences
two forms of hot spots, one due to the flow maldistribution within the system and the other due to the additional
non-uniform heat generation by the device. The hot spots have been shown to exhibit drastically different shapes
and core temperatures and this has been verified through simulations and infrared thermography. To efficiently
cool hot spot core temperatures, nanofluids have been employed and ‘smart cooling’ has been observed and the
same has been explained based on nanoparticle slip mechanisms. The present work shows that the notion that
high flow maldistribution leads to high thermal maldistribution, is not always true and existing maldistribution
can be effectively utilized to tackle specific hot spot location. The present work can be important to design
cooling mechanisms for real microprocessors with high core activity leading to non-uniform hot spot formation
probabilities.
