| CPC G06F 1/20 (2013.01) [H01L 2225/06541 (2013.01); H01L 2225/06589 (2013.01)] | 20 Claims |

|
1. A method for fast prediction of hotspot temperatures in three-dimensional (3D) integrated circuit (IC) chips, comprising:
inputting a dataset comprising thermal simulation data and associated parameters for 3D IC chips, wherein the dataset includes variables associated with power distribution, processor core location distribution, and Through Silicon Via (TSV) distribution;
applying a K-fold Cross Validation (K-CV) algorithm to train a machine learning model using the inputted dataset, wherein the K-CV algorithm is employed to optimize the training process and prevent overfitting; and
utilizing a Support Vector Regression (SVR) algorithm within the trained machine learning model to predict hotspot temperatures of the 3D IC chips; and
adjusting a cooling strategy for cooling the 3D IC chips, in response to monitoring the hotspot temperatures via the trained machine learning model.
|