首页> 外文会议>IEEE Intersociety Conference on Thermal and Thermomechanical Phenomena in Electronic Systems >Cold Plate Pin-Fin Optimization for Multi-Die Systems Using Design of Experiment
【24h】

Cold Plate Pin-Fin Optimization for Multi-Die Systems Using Design of Experiment

机译:基于实验设计的多芯片系统冷板针脚优化

获取原文

摘要

In the era of “Internet of everything (IoE)”, humans are connected to everything by internet which has rapidly risen the demand for powerful mobile devices as well as data centers with high performance processors. High speed and high-quality data processing and data transfer between heterogeneous dies in one integrated system are inevitably required. 2.5D stacked integrated circuits (2.5D-SICs) with through-silicon vias (TSV), as next generation silicon technologies are capable of providing desired multi-functional performance. Both high power heterogeneous dies, and limited space in a compact system contribute to the thermal bottleneck of 2.5D-SICs. Microfluidic cooling with non-uniform pin-fin is a promising approach to provide sufficient non-uniform cooling capacity. Previous work demonstrated that pin-fin enhanced micro-gap cooling can address the heat dissipation challenge for both 2.5D and 3D-SICs. In this paper, five heat sources, including one field programmable gate array (FPGA) of 25 mmx25 mm, and four heaters, each 6 mmx6 mm, are adopted in a thermal test vehicle. Non-uniform pin-fin structures for five dies in the micro-gap cold plate have been systematically optimized by utilizing design of experiment (DoE) with full-scale computational fluid dynamics/heat transfer (CFD/HT) simulations. With the guidance of DoE, representative cases with various pin-fin structures have been simulated by CFD/HT. Optimized both pin and fin structures for dies have been proposed efficiently. These optimized structures in the cold plate for multi-die system have been verified by CFD/ HT, which have met the target temperatures of five dies with lowest pressure drop. Methods of design of experiment (DoE) and CFD/HT can be utilized together to achieve multi-component system optimization in customized ranges.
机译:在“万物互联(IoE)”时代,人类通过互联网连接到万物,这对功能强大的移动设备以及具有高性能处理器的数据中心的需求迅速增长。不可避免地需要在一个集成系统中进行高速,高质量的数据处理以及异构芯片之间的数据传输。具有直通硅通孔(TSV)的2.5D堆叠集成电路(2.5D-SIC),因为下一代硅技术能够提供所需的多功能性能。高功率异质芯片和紧凑系统中有限的空间都造成了2.5D-SIC的热瓶颈。具有不均匀销钉鳍的微流体冷却是一种有前途的方法,可以提供足够的不均匀冷却能力。先前的工作表明,针鳍增强的微间隙冷却可以解决2.5D和3D-SIC的散热难题。在本文中,在热测试车中采用了五个热源,包括一个25mmx25mm的现场可编程门阵列(FPGA)和四个每个6mmx6mm的加热器。通过利用实验设计(DoE)和全面的计算流体动力学/热传递(CFD / HT)模拟,系统地优化了微间隙冷板中五个模具的非均匀针翅结构。在能源部的指导下,用CFD / HT模拟了具有各种针鳍结构的典型案例。已经有效地提出了用于模具的优化的销和鳍结构。通过CFD / HT验证了用于多模具系统的冷板中的这些优化结构,它们满足了具有最低压降的五个模具的目标温度。实验设计(DoE)和CFD / HT的方法可以一起使用,以实现自定义范围内的多组件系统优化。

著录项

相似文献

  • 外文文献
  • 中文文献
  • 专利
获取原文

客服邮箱:kefu@zhangqiaokeyan.com

京公网安备:11010802029741号 ICP备案号:京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有
  • 客服微信

  • 服务号