FPGARay: Accelerating Physically Based Rendering Using FPGAs

Information

Abstract

The synthesis of an image from a scene stored on a computer is called rendering, which is able to deliver photo-realistic results, e.g., by using specific variants of the class of ray tracing algorithms. However, these variants (e.g., path tracing) possess a stochastic characteristic which results in a high computational expense. This is explained by the nature of stochastic algorithms, which use a high number of samples to compute a result—in case of ray tracing, these samples manifest in a high number of rays needed for a complete rendering.

One possibility to accelerate ray tracing—no matter if using a stochastic or simpler variants—is the use of customized hardware. FPGRay is such an approach, which combines the use of customized hardware with the software of an off-the-shelf PC to a hybrid solution. This allows increasing the efficiency by specialized hardware and delivers a sustainability in case of changing algorithms at the same time.

The results point towards a possible efficiency gain. Unfortunately, in the scope of this thesis this was not realizable and the specific implementation showed a lower efficiency compared to the software implementation. Nevertheless, the possibility to achieve a higher efficiency with this approach by indicating FPGRay’s potential could be shown.

Additional Files and Images

No additional files or images.

Weblinks

No further information available.

BibTeX

@mastersthesis{reznicek-2020-fpgaray,
  title =      "FPGARay: Accelerating Physically Based Rendering Using FPGAs",
  author =     "Alexander Reznicek",
  year =       "2020",
  abstract =   "The synthesis of an image from a scene stored on a computer
               is called rendering, which is able to deliver
               photo-realistic results, e.g., by using specific variants of
               the class of ray tracing algorithms. However, these variants
               (e.g., path tracing) possess a stochastic characteristic
               which results in a high computational expense. This is
               explained by the nature of stochastic algorithms, which use
               a high number of samples to compute a result—in case of
               ray tracing, these samples manifest in a high number of rays
               needed for a complete rendering.  One possibility to
               accelerate ray tracing—no matter if using a stochastic or
               simpler variants—is the use of customized hardware. FPGRay
               is such an approach, which combines the use of customized
               hardware with the software of an off-the-shelf PC to a
               hybrid solution. This allows increasing the efficiency by
               specialized hardware and delivers a sustainability in case
               of changing algorithms at the same time.  The results point
               towards a possible efficiency gain. Unfortunately, in the
               scope of this thesis this was not realizable and the
               specific implementation showed a lower efficiency compared
               to the software implementation. Nevertheless, the
               possibility to achieve a higher efficiency with this
               approach by indicating FPGRay’s potential could be shown.",
  month =      jun,
  address =    "Favoritenstrasse 9-11/E193-02, A-1040 Vienna, Austria",
  school =     "Research Unit of Computer Graphics, Institute of Visual
               Computing and Human-Centered Technology, Faculty of
               Informatics, TU Wien ",
  URL =        "https://www.cg.tuwien.ac.at/research/publications/2020/reznicek-2020-fpgaray/",
}