Mehdi Saeidi

Thermal implications of mobile 3D-ICs

In this paper, we introduce a fast and accurate 3D thermal analysis methodology for sequential face-to-back (F2B) and parallel face-to-face (F2F) 3D integration technologies. Our proposed models take design floorplan and the corresponding power map with temperature-dependent leakage power component and accurately estimate the temperature of any given location in the design. We show that our models are within less than 1% of commercial-grade finite-volume models while they are 100X faster. In addition, we demonstrate the minimum power reduction required in mobile 3D-ICs to achieve similar thermal behaviour of that of the corresponding 2D-ICs. Subsequently, we provide guidelines on 3D partitioning / floorplanning to further improve thermal profile of 3D-ICs.

Adaptive Regression-Based Thermal Modeling and Optimization for Monolithic 3-D ICs

In this paper, we first present a comprehensive study of the unique thermal behavior in monolithic 3-D integrated circuits (ICs) in contrast to through silicon via-based 3-D ICs. In particular, we study the impact of the thin interlayer dielectric between the device tiers on vertical thermal coupling. We then study and compare the impact of different application-based package structures on the thermal behavior of monolithic 3-D ICs. With these unique properties and behavior, we develop a fast and accurate compact full-chip thermal analysis model based on nonlinear regression technique which adapts to the package structure during development and hence considers it during temperature evaluation. Our model is extremely fast and highly accurate with an error of less than 5%. This model is incorporated into a thermal-aware 3-D-floorplanner that runs without significant runtime overhead. We use the floorplanner with our package-aware thermal model and observe up to 22% reduction in the maximum temperature with insignificant area and performance overhead.

Fast and accurate thermal analysis of smartphone with dynamic power management using reduced order modeling

Power saving techniques and associated thermal management are inevitable for present day smartphones. Smartphones employ thermal feedback (temperature at key locations) based dynamic power management strategies to maintain the system temperature in the desired range. It is challenging to simulate the thermal behavior of such a system as running a computational fluid dynamics system with control logic consumes significant time and computation resources. An efficient transient thermal model for smartphone simulation based on Linear Time Invariant (LTI) system is proposed in this paper. State space model developed based on LTI system can be used for transient thermal simulations with similar accuracy as full 3D transient CFD model but a significantly faster run time. The model generation process starts with computation fluid dynamics (CFD) results of a smartphone model. A state model is created which is very efficient and runs orders of magnitude faster. Extracted state space model can be used to check different variations of power control logics for a given smart phone design without the need to perform Full CFD analysis. In this paper, case study has been conducted to compare results from state space model with full CFD model for specific control logic.

CTSIM: Convolution-based thermal simulation using iterative methods

Mobile processors push the envelope of thermal design due to lack of active cooling and heavy computational requirements. Many different use case applications must be analyzed to understand the thermal risks involved including the device leakage power, which has an exponential dependence on temperature. Commercial computational fluid dynamic (CFD) solvers generally take more than four hours for a single smartphone simulation with acceptable accuracy without accounting the for the leakage power. In this paper, CTSIM is presented which is a compact thermal solver (CTS) which uses convolution and iterative methods. CTSIM is as accurate as commercial solvers with a significant speed improvement in repeated simulation time for use case and benchmark analysis. Additionally, the temperature dependence on leakage is also accounted for correctly. The result is a fast and compact thermal model which provides commercial CFD accurate analyses with an 8000x speed improvement.