logo80lv
Articlesclick_arrow
Research
Talentsclick_arrow
Events
Workshops
Aboutclick_arrow
profile_loginLogIn

Learn More about Technology Behind Pixar's RenderMan XPU

See how it utilizes both CPU and GPU.

Pixar

Pixar's famous RenderMan renderer has evolved with time, and now, the company has RenderMan XPU, a rewritten version of the renderer, designed to run on both CPUs and GPUs.

You can find its technical details from the paper called RenderMan XPU: A Hybrid CPU+GPU Renderer for Interactive and Final-Frame Rendering.

The first version of RenderMan was based on the Reyes algorithm and used to render the first CG animated feature film, Toy Story, and many other movies. Pixar later augmented the Reyes algorithm with multithreading, ray-traced shadows and reflections, global illumination, level-of-detail tessellation, subsurface scattering, and fast point-based approximations for better results.

With RenderMan XPU, the key abilities for the design were high-performance rendering on heterogeneous hardware, a modular architecture for flexibility and scalability, and advanced rendering features for high-quality images.

As the company reports, this new version is up to 2.3 times faster than the previous version on CPUs, up to 10 times on GPUs, and 15 times on both.

"XPU uses a wavefront-based design based on tracing a “wave” of rays over the scene, as opposed to a design based on megakernels (where individual ray paths are processed in isolation)."

The light transportation is simulated with an integrator that implements a unidirectional Monte Carlo path tracing algorithm, "iterating over light reflection and refraction bounces “backwards” from the camera to the light sources."

"RenderMan XPU is a single renderer with a dual purpose: interactive rendering for fast feedback, and off-line rendering of movie-quality final frames. It utilizes heterogeneous hardware for optimal use of available compute power and memory capacity, within tools and workflows that artists are used to. It has a modular and flexible architecture, and source code is shared as much as possible across different types of hardware."

Read more in the full version of the paper and join our 80 Level Talent platform and our new Discord server, follow us on InstagramTwitterLinkedInTelegramTikTok, and Threads, where we share breakdowns, the latest news, awesome artworks, and more.

Join discussion

Comments 0

    You might also like

    We need your consent

    We use cookies on this website to make your browsing experience better. By using the site you agree to our use of cookies.Learn more