SHF: Small: Visual Architectures: Engaging Crowds in Design and Discovery for Custom Reconfigurable Devices

定制的可重构计算平台提供低功耗、高性能、高效率以及针对特定应用领域的需求量身定制的增加的冗余。成功创建这些高度定制的特定领域架构的能力提供了巨大的优势，包括数量级的功耗节省，更长的电池寿命，更小，更快，更强大的设备，以及更短的上市时间。然而，使极端定制成为设计过程中不可或缺的一部分，需要设计更加简单，以创建直接满足特定领域需求的新颖的开箱即用架构。这一挑战可以分为三个具体领域：（1）发现快速，高效的算法，允许快速设计空间探索;（2）通过促进非科学家和非工程师的计算思维来增加参与并扩大社区;（3）通过创新的视觉框架教育下一代定制芯片设计师。有效的多人游戏和合作游戏将在这个项目中开发，允许玩家团队合作解决更大范围和复杂性的映射和设计问题。在这方面，该项目还主要涉及代表性不足的群体的成员。这项研究是围绕着开发视觉架构，以解决上述挑战的想法。更具体地说，该项目涉及：（a）创建一个高度可视化、类似游戏的设计环境，从而使复杂的架构和映射问题变得容易理解，从而提高参与度并显著增加参与度;（B）分析通过不断扩大的玩家社区产生的大量数据，以发现更好、更有效的自动算法;（c）利用由此建立的互动平台，教育下一代芯片设计师，培养他们的技能和直觉。

Custom reconfigurable computing platforms offer low power, high performance, high efficiency, and increased redundancy tailored to the needs of a specific application domain. The ability to successfully create these highly customized domain-specific architectures offers tremendous advantages, including orders of magnitude power savings, longer battery lives, smaller, faster, more robust devices, and a shorter time to market. However, making extreme customization an integral part of the design process requires design to be significantly simpler to create novel, out of the box architectures that directly address the requirements of a specific domain. This challenge can be divided into three specific areas: (1) discover fast, efficient algorithms that allow for rapid design space exploration; (2) increase participation and widen the community by promoting computational thinking among non-scientists and non-engineers; and (3) educate the next generation of custom chip designers through innovative visual frameworks. Effective multiplayer and cooperative games will be developed in this project that allow teams of players to collaboratively solve mapping and design problems of much larger scope and complexity. In this respect the project also involves members if underrepresented groups in a major way. This research is centered around the idea of developing Visual Architectures to address the challenges outlined above. More specifically, the project involves: (a) creating a highly visual, game-like design environment, thus making complex architecture and mapping problems accessible leading to greater engagement and significantly increasing participation; (b) analyze the vast amounts of data generated through the widening player community to discover better, more efficient automated algorithms; (c) employ the interactive platforms thus created to educate the next generation of chip designers and build their skills and intuitions.