NanoLeak: A Fast Analytical Green’s Function-based Leakage-aware Thermal Simulator

In this paper, we propose NanoLeak, a comprehensive temperature simulator that incorporates both classical heat transfer mechanisms and nanoscale effects. It performs both steady state and transient analyses while automatically taking leakage into account. We derive closed-form expressions for the Green’s function (impulse response of a power source) for all scenarios with leakage; there is no need for the traditional, time-consuming iterative solutions. We show a speedup of 1250X for the classical heat transfer case (Fourier’s heat equation) with an error limited to 2.4% in computing the steady-state thermal profile. At nanoscale levels, the Boltzmann transport equation (BTE) is solved to analyze the temperature profile. In this paper, we analytically compute the leakage-aware solution for the gray-BTE model and compare the results against competing, state-of-the-art work (211-2580X speedup).