TY - GEN
T1 - PV-Clock
T2 - 28th IEEE Computer Society Annual Symposium on VLSI, ISVLSI 2025
AU - Wang, Yiyu
AU - Pavlidis, Vasilis F.
AU - Wang, Rui
AU - Cheng, Yuanqing
N1 - Publisher Copyright:
© 2025 IEEE.
PY - 2025
Y1 - 2025
N2 - Three-dimensional integration technology enables higher density and heterogeneous integration, extending Moore's Law. However, increasing integration density complicates clock network design. Existing synthesis methods often neglect process variations, which are inherent in fabrication and degrade clock signal precision. Additionally, optimizing one design metric, such as reducing clock skew, often increases power consumption, limiting overall efficiency. These trade-offs restrict the design space and hinder 3D IC performance. To address these challenges, this paper proposes a novel process variation-aware clock network synthesis technique incorporating a multi-objective collaborative optimization framework. The proposed method effectively balances clock skew and power consumption while enhancing robustness. Experimental results demonstrate significant improvements: clock skew is reduced from 21.54 p s to 15.02 p s, while clock power decreases by 23.8%, highlighting the method's effectiveness in mitigating process variations and optimizing 3D clock networks.
AB - Three-dimensional integration technology enables higher density and heterogeneous integration, extending Moore's Law. However, increasing integration density complicates clock network design. Existing synthesis methods often neglect process variations, which are inherent in fabrication and degrade clock signal precision. Additionally, optimizing one design metric, such as reducing clock skew, often increases power consumption, limiting overall efficiency. These trade-offs restrict the design space and hinder 3D IC performance. To address these challenges, this paper proposes a novel process variation-aware clock network synthesis technique incorporating a multi-objective collaborative optimization framework. The proposed method effectively balances clock skew and power consumption while enhancing robustness. Experimental results demonstrate significant improvements: clock skew is reduced from 21.54 p s to 15.02 p s, while clock power decreases by 23.8%, highlighting the method's effectiveness in mitigating process variations and optimizing 3D clock networks.
KW - 3D ICs
KW - Clock Network Synthesis
KW - Geometric Optimization
KW - Process Variations
UR - https://www.scopus.com/pages/publications/105016145068
U2 - 10.1109/ISVLSI65124.2025.11130256
DO - 10.1109/ISVLSI65124.2025.11130256
M3 - 会议稿件
AN - SCOPUS:105016145068
T3 - Proceedings of IEEE Computer Society Annual Symposium on VLSI, ISVLSI
BT - IEEE Computer Society Annual Symposium on VLSI, ISVLSI 2025 - Conference Proceedings
PB - IEEE Computer Society
Y2 - 6 July 2025 through 9 July 2025
ER -