GPU Evolution-Inspired Optimization of Heat Sinks for Practical Manufacturing Processes

Heat sinks are typically designed by hand. Recently, computational methods have been applied to optimize the shape of heat sinks to maximize their effectiveness. However, these designs are typically impractical for real-world manufacturing. This paper describes an evolution-inspired method accelerated by GPU for automatically designing heat sinks within the constraints of common manufacturing processes. The resulting heat sink designs are presented.

---

Contributors:

• Liwen Shih (U of Houston - Clear Lake)
• Daniel J. Williams (U of Houston - Clear Lake)