Enhancing the Power of Polyhedral-Based Optimizations with Coordinate-Based Hill Climbing
本文通过在多面体编译器Pluto中加入基于坐标的爬山调优方法，调整数字转换参数以优化内核性能，解决了传统优化方法难以达到最优解的问题。
0 citationsRead paper
Selected work
Representative Papers
DeepOHeat-v2: Self-Improving Operator Learning for Fast and Trustworthy Thermal Optimization in 3D-IC Design
This work proposes DeepOHeat-v2 to address training failures in operator learning for 3D-IC thermal optimization caused by material discontinuities and ill-conditioning. The method employs a discrete energy-based physics loss and a matrix-preconditioned optimizer to overcome interface discontinuities and numerical instability, while introducing an incremental self-improvement framework with hotspot trust gating to enhance generalization. Experimental results demonstrate that DeepOHeat-v2 achieves a peak temperature error of 0.55 K and reduces design temperature discrepancy to 0.11 K. By matching traditional solver accuracy while delivering a 56× speedup, this approach effectively overcomes thermal simulation bottlenecks in high-contrast multi-chip stacking scenarios, offering a robust and efficient solution for advanced 3D-IC thermal analysis.
0 citationsRead paper
Vanilla Object Orientation (VOO): A Value-Semantics Approach to Classes in Tcl
This work addresses the high overhead, poor memory efficiency, and complex destruction semantics inherent in traditional Tcl object systems such as TclOO and Itcl, which rely on external frameworks. The authors propose a lightweight object-oriented framework built exclusively on native Tcl lists and dictionaries, introducing for the first time in Tcl a pure value-based semantics for classes and objects. By leveraging copy-on-write mechanisms, the approach enables automatic memory management without requiring explicit destructors or additional runtime support. The design preserves standard Tcl calling conventions and offers a compatible migration path to C++. Experimental results demonstrate that, on Tcl 8.6.13 and 9.0, object creation is 7–18 times faster and memory efficiency improves by 4–6 times; the corresponding C++ implementation achieves 2.3–2.6 times faster field access and reduces memory consumption by 6.8–9.8 times.
0 citationsRead paper
Recent publications
Latest Papers
Enhancing the Power of Polyhedral-Based Optimizations with Coordinate-Based Hill Climbing
本文通过在多面体编译器Pluto中加入基于坐标的爬山调优方法，调整数字转换参数以优化内核性能，解决了传统优化方法难以达到最优解的问题。
0 citationsRead paper
DeepOHeat-v2: Self-Improving Operator Learning for Fast and Trustworthy Thermal Optimization in 3D-IC Design
This work proposes DeepOHeat-v2 to address training failures in operator learning for 3D-IC thermal optimization caused by material discontinuities and ill-conditioning. The method employs a discrete energy-based physics loss and a matrix-preconditioned optimizer to overcome interface discontinuities and numerical instability, while introducing an incremental self-improvement framework with hotspot trust gating to enhance generalization. Experimental results demonstrate that DeepOHeat-v2 achieves a peak temperature error of 0.55 K and reduces design temperature discrepancy to 0.11 K. By matching traditional solver accuracy while delivering a 56× speedup, this approach effectively overcomes thermal simulation bottlenecks in high-contrast multi-chip stacking scenarios, offering a robust and efficient solution for advanced 3D-IC thermal analysis.
0 citationsRead paper
Vanilla Object Orientation (VOO): A Value-Semantics Approach to Classes in Tcl
This work addresses the high overhead, poor memory efficiency, and complex destruction semantics inherent in traditional Tcl object systems such as TclOO and Itcl, which rely on external frameworks. The authors propose a lightweight object-oriented framework built exclusively on native Tcl lists and dictionaries, introducing for the first time in Tcl a pure value-based semantics for classes and objects. By leveraging copy-on-write mechanisms, the approach enables automatic memory management without requiring explicit destructors or additional runtime support. The design preserves standard Tcl calling conventions and offers a compatible migration path to C++. Experimental results demonstrate that, on Tcl 8.6.13 and 9.0, object creation is 7–18 times faster and memory efficiency improves by 4–6 times; the corresponding C++ implementation achieves 2.3–2.6 times faster field access and reduces memory consumption by 6.8–9.8 times.
0 citationsRead paper