Fast High-Resolution Thermal Simulation for Integrated Circuits Using Multi-Block Proper Orthogonal Decomposition
演講摘要:
A methodology based on multi-block Proper Orthogonal Decomposition (POD) will be presented for thermal simulation of a physical structure with a high degree of geometric repetition. The proposed approach projects the problem onto a functional space in order to reduce the numerical degrees of freedom (DOF’s). The approach has been applied to thermal simulations of IC’s subjected to power pulses initiated by digital signals. It has been demonstrated that the POD model offers accurate thermal solution as detailed as numerical simulation and is able to reduce numerical DOF’s by nearly 6 orders of magnitude in 3D structures.
A methodology based on multi-block Proper Orthogonal Decomposition (POD) will be presented for thermal simulation of a physical structure with a high degree of geometric repetition. The proposed approach projects the problem onto a functional space in order to reduce the numerical degrees of freedom (DOF’s). The approach has been applied to thermal simulations of IC’s subjected to power pulses initiated by digital signals. It has been demonstrated that the POD model offers accurate thermal solution as detailed as numerical simulation and is able to reduce numerical DOF’s by nearly 6 orders of magnitude in 3D structures.
場次:
3
演講日期:
2017-03-10
主講人:
鄭明正 教授 (Department of Electrical & Computer Engineering, Clarkson University)
瀏覽數:
分享






