RadeonProRenderSDK
AMD Radeon™ ProRender is a powerful physically-based path traced rendering engine that enables creative professionals to produce stunningly photorealistic images.
Install / Use
/learn @GPUOpen-LibrariesAndSDKs/RadeonProRenderSDKREADME
AMD Radeon:tm: ProRender SDK

AMD Radeon:tm: ProRender is a powerful physically-based rendering engine that enables creative professionals to produce stunningly photorealistic images. https://www.amd.com/en/technologies/radeon-prorender
This repo contains :
-
The AMD Radeon:tm: ProRender library: includes and built binaries.
Install Precompiled Kernels
Since 3.01.00, for Northstar the default render backend is HIP ( instead of OpenCL ).
The main difference is that compute kernels are now precompiled by us. They must be downloaded from a submodule.
Note that for now you can still use the OpenCL backend ( with RPR_CREATION_FLAGS_ENABLE_OPENCL in rprCreateContext ). However we don't recommend it as in the future we may put less resource to support this backend.
Download the precompiled kernels with the command:
git submodule update --init --recursive
The precompiled kernels folder can be modified with RPR_CONTEXT_PRECOMPILED_BINARY_PATH. ( for its usage, check the tutorials ).
If precompiled kernels are not found,RPR_ERROR_SHADER_COMPILATION is returned by rprContextRender.
Building and running the Tutorials
On Visual Studio:
> cd tutorials
> ..\premake5\win\premake5.exe vs2022
( replace vs2022 by the version of your choice: vs2019, vs2017... )
then open tutorials/Tutorials.sln
On Ubuntu 20:
Dependencies on Linux for the tutorials: GLEW, GLUT, Pthread.
> cd tutorials
> sudo chmod +x ../premake5/linux64/premake5
> ../premake5/linux64/premake5 gmake
> make -j config=release_x64
On Centos 7:
Same dependencies than Ubuntu.
You should also update your GCC, we have tested it with devtoolset-10 (yum install devtoolset-10)
> cd tutorials
> sudo chmod +x ../premake5/linux64/premake5
> ../premake5/linux64/premake5 --centos gmake
> source scl_source enable devtoolset-10
> make -j config=release_x64
On MacOS:
> cd tutorials
> sudo chmod +x ../premake5/osx/premake5
> ../premake5/osx/premake5 gmake
> make -j config=release_x64
To run a demo, just select the executable and start it, example:
> cd tutorials/Bin
> ./05_basic_scene64
this demo generates the rendering output images: 05_00.png, 05_01.png, 05_02.png, 05_03.png
See detailed documentation at https://radeon-pro.github.io/RadeonProRenderDocs/sdk/tutorials.html
Python Binding
You can create a RPR Python binding extension.
First, make sure to have all the submodules recursively ( Nanobind is used to create the binding ):
git submodule update --init --recursive
Build the RPR binding libraries:
> cd python/
> mkdir build
> cd build
> cmake -Dnanobind_DIR=nanobind/cmake ..
> cmake --build . --config Release
Those builds will be generated inside python/build/ they are needed by the python script running RPR in order to execute the:
import rpr
import rprs
import rprgltf
Note that if you have several pythons installed on your system, you may need to define Python_EXECUTABLE in the first cmake instruction. Example:
cmake -Dnanobind_DIR=nanobind/cmake -DPython_EXECUTABLE=/usr/bin/python3 ..
The name of the compatible python version can be found in the name of the generated file. for example rpr.cpython-38-x86_64-linux-gnu.so means it's for python 3.8.
Run some test scripts:
> cd python/test
> python test_script_rpr.py
> python test_script_gltf.py
If tests run correctly, you should find the generated rendering images inside the python/test folder.
Additional Info
-
Radeon:tm: ProRender GPUOpen web site https://gpuopen.com/radeon-pro-render/
-
Radeon:tm: ProRenderSDK documentation https://radeon-pro.github.io/RadeonProRenderDocs/sdk/about.html
Related Skills
node-connect
345.9kDiagnose OpenClaw node connection and pairing failures for Android, iOS, and macOS companion apps
frontend-design
106.4kCreate distinctive, production-grade frontend interfaces with high design quality. Use this skill when the user asks to build web components, pages, or applications. Generates creative, polished code that avoids generic AI aesthetics.
openai-whisper-api
345.9kTranscribe audio via OpenAI Audio Transcriptions API (Whisper).
qqbot-media
345.9kQQBot 富媒体收发能力。使用 <qqmedia> 标签,系统根据文件扩展名自动识别类型(图片/语音/视频/文件)。

