An updated version of an accelerated CUDA ray-tracer I started at https://github.com/rogerallen/raytracinginoneweekendincuda based on the first book in the https://raytracing.github.io/ series.

I started this in 2018 and set it aside until 2022 when I taught a class at work and modernized the code to match changes at https://raytracing.github.io/

This is still just a toy renderer and just something I'm working on for personal enjoyment. Everyone should build their own ray tracer.

You will need to set the options for the CUDA architecture to compile for in the NVCC_GENCODE env variable. See https://arnon.dk/matching-sm-architectures-arch-and-gencode-for-various-nvidia-cards/

export NVCC_GENCODE="-arch sm_61 -gencode=arch=compute_61,code=sm_61"
make
make scenes/test1.png scenes/test1d.png scenes/test1c.png

If you get CUDA error = 209 then you have a mismatch between the compiled kernel architecture & your actual GPU architecture.

Additionally, you'll want to set the compiler option -DCOMPILING_FOR_WSL if you are compiling on Windows Subsystem for Linux. I just add that to the NVCC_GENCODE variable.

There are 3 executables:

• rrt is the CUDA accelerated code using float floating point
• rrtd is the CUDA accelerated code using double floating point
• rrtc is the less accelerated C++ code
• rrto is the CPU/OpenMP accelerated C++ code

Usage: rrt [options]
  -i file.txt         : input scene file
  -o file.png         : output raytraced PNG image (default is PPM to stdout)
  -w <width>          : output image width. (default = 1200)
  -h <height>         : output image height. (800)
  -s <samples>        : number of samples per pixel. (10)
  -d <max_depth>      : may ray recursion depth. (50)
  -tx <num_threads_x> : number of threads per block in x. (8)
  -ty <num_threads_y> : number of threads per block in y. (8)
  -q                  : query devices & cuda info
  -D <device number>  : use this cuda device (0)

Some updates I made that are my own improvements from the original book.

• added commandline arguments
• added PNG output (via stb_image_write.h)
• use float or double math in renderer
• add a scene description file
• added objects based on triangles

Updates from "The Next Week"

• added motion blur & moving spheres

• add motion blur
• add motion blur to object instances
• add BVH
• input list of scenes to render
• cmake build?
• add pybind11 to enable running from python

To the extent possible under law, Roger Allen has waived all copyright and related or neighboring rights to Roger's Ray Tracer.