Hardware Support for Interactive Ray Tracing

犹他州硬件光线跟踪小组正在设计使用光线跟踪来生成计算机图形图像的特殊用途硬件。光线跟踪创建比现有商业图形芯片更高质量和更逼真的图像，所有这些都使用基于z缓冲区的算法，并且在场景对象数据库非常大或基本体对象不是简单三角形的情况下效率更高。然而，光线跟踪被认为速度太慢，无法在商用计算机上交互工作。这项研究的最终目标是开发能够实现商用光线跟踪芯片的技术，这种芯片可以取代今天的图形处理单元，并为未来的计算机提供更高质量和更逼真的图形能力。这项研究将光线跟踪算法领域投射到包括多条光线跟踪管道的专用处理器中。光线跟踪是一种自然的并行应用程序，可以利用并行硬件，而处理器硬件的专门化使性能大大提高，而不是运行在通用CPU上的软件。他们的设计策略基于之前在设计特定于域的处理器时所做的工作，其中包括数据路径的运行时配置。这允许该设备以非常接近完全定制流水线的速度运行，但具有足够的可编程性，以便在制造后能够适应算法的变化。虽然专用硬件和光线跟踪是活跃的研究领域，但对于专用光线跟踪硬件的设计却知之甚少。该项目包括一名VLSI研究人员、一名存储器架构师和两名光线跟踪专家，这正是该项目所需的领域。该项目将产生与计算机图形以及特殊用途存储系统和计算流水线设计相关的丰富知识。该项目的成果有可能帮助将高端计算机图形功能移植到每台PC上。该组织的最终目标是将目前只有计算精英才能使用的图形功能带入日常使用。

The Utah Hardware Ray Tracing group is designing special purpose hardware that uses ray tracing to generate computer graphic images. Ray tracing creates much higher quality and more realistic images than existing commercial graphics chips, all of which use a z-buffer based algorithm, and is more efficient when the scene object database is extremely large or where the primitive objects are not simple triangles. However, ray tracing is considered too slow to work interactively on commodity computers. The end goal of this research is to develop technology to enable commodity ray trace chips that can replace today's graphics processing units and enable higher quality and more realistic graphics capabilities for future computers. This research casts the ray tracing algorithm domain into a specialized processor that includes multiple ray tracing pipelines. Ray tracing is a naturally parallel application that can take advantage of parallel hardware, and specialization of the processor hardware enables vastly increased performance over software running on a generic CPU. Their design strategy is based on previous work in designing domain specific processors that includes run-time configuration of the datapath. This allows the device to operate at very close to the speed of a fully custom pipeline, but with enough programmability so that variations in algorithm can be accommodated after fabrication. While special purpose hardware and ray tracing are active research areas, little is known about the design of special purpose ray tracing hardware. This project includes a VLSI researcher, a memory architect, and two ray tracing specialists, precisely the areas needed for this project. The project will produce knowledge about a rich set of intellectual issues related to computer graphics and the design of special purpose memory systems and computational pipelines. The results of this project have the potential to help move high-end computer graphics capabilities onto every PC. The group's end goal is to bring a graphics capability that is currently only available to the computing elite into everyday use.