Performance Modeling of Multicore Processors with Process Variations


Amdahls law predicts limits to performance speedup based on available parallelism in different segments of the application. We have extended Amdahls law for modelling multicores with process variation using simple mathematical techniques. We have proposed three separate corollaries to the standard Amdahls law to model the performance of different multicore configurations with different modes of operation.

(Published in International Journal of High Performance Systems Architecture (IJHPSA ) 2013)

Leakage Aware Ultrafast Temperature Simulator

On-chip temperature is increasingly being regarded as a first class design constraint. Temperature has a direct effect on the amount of leakage power, and lifetime reliability. We have proposed the design of an ultra-fast thermal simulator that can perform both steady state and transient thermal analysis, and also take the effect of leakage power into account. We have used a novel Hankel transform based technique to derive a transient version of the Greens function for a chip, which takes into account the feedback loop between temperature and leakage.

(Published in Asia and South Pacific Design Automation Conference (ASP-DAC ) 2014)

Power Aware Task Assignment Algorithms

We have demonstrated the implications of lateral heat conduction on leakage power consumption and showed that heat-spread aware task assignment can significantly impact the total power consumption. In this context, We have developed an application mapping framework LeakOpt, that minimizes the total power consumption of manycore processors.

(Published in  IEEE Computer Society Annual Symposium on VLSI (ISVLSI),Pittsburgh, USA 2016)