Skip to main content

Intel OSPRay

v3.2.0

High-performance CPU ray tracing framework for scientific visualization and rendering

Ray TracingVolume
C++
Apache-2.0
Active
CPU
Stars
1.0k
Latest Release3.2.0
Release DateSep 2024
Contributors40
Forks180
At a Glance
Technique
Ray Tracing, Volume
Language
C++
License
Apache-2.0
Platforms
Linux
macOS
Windows
GPU Support
No
CPU Support
Yes
Scene Formats
OBJ, glTF, GLB
Output Formats
PNG, JPEG, PPM, EXR
First Release
Jan 2015
Latest Release
3.2.0 — Sep 2024
Best For
High-quality CPU ray tracing for scientific visualization, particularly on HPC clusters without GPU hardware, and as a rendering backend for VTK and ParaView

Development Activity

Commit activity data is not available for this renderer.

1.0k
Stars
3.2.0
1 year ago
40
Contributors
View on GitHub

Overview

Best for

High-quality CPU ray tracing for scientific visualization, particularly on HPC clusters without GPU hardware, and as a rendering backend for VTK and ParaView

Not ideal for

GPU-accelerated rendering, real-time game development, artists wanting a GUI-based workflow, or applications preferring GPU ray tracing via OptiX or DXR

Strengths

  • Achieves high-quality ray-traced rendering entirely on CPU via ISPC vectorization and Embree, critical for HPC clusters where GPUs are unavailable or where datasets exceed GPU memory
  • MPI-parallel distributed rendering scales across hundreds of nodes, enabling visualization of petascale datasets that no single machine could hold in memory
  • Two-renderer architecture (SciVis + Path Tracer) allows users to trade speed for quality — fast ambient occlusion for interactive exploration or full path tracing for publication imagery
  • Deep integration with VTK and ParaView allows scientists to switch to ray-traced rendering without changing their visualization pipeline — just select the OSPRay backend
  • Intel Open Image Denoise integration provides AI-driven denoising that dramatically reduces the sample count needed for clean path-traced images

Limitations

  • CPU-only by design — cannot leverage GPU hardware for rendering, even when high-end GPUs are available on the system
  • Performance is optimized for Intel architectures via ISPC targeting SSE, AVX, and AVX-512 — functional on AMD CPUs but may not achieve peak performance
  • Not a standalone application — requires integration into a host application (VTK, ParaView, custom code) or use of example viewers for rendering
  • Lower-level C API requires more development effort than higher-level tools — most users interact via VTK or ParaView rather than programming directly against OSPRay
  • Material system in SciVis mode is basic compared to production renderers — the path tracer supports PBR but with fewer material types than Mitsuba or Cycles

Background

Intel OSPRay is an open source, scalable, and portable CPU ray tracing framework designed for high-fidelity scientific visualization. Part of Intel's oneAPI Rendering Toolkit, OSPRay builds on Intel Embree for ray tracing acceleration and Intel Open Volume Kernel Library (OpenVKL) for volume rendering, delivering performance levels that typically require GPU hardware while running entirely on CPU — a critical advantage for HPC clusters where GPU availability is limited or where datasets exceed GPU memory capacity.

OSPRay provides two primary renderers: a SciVis renderer optimized for scientific visualization with ambient occlusion, shadows, and efficient basic shading, and a physically based path tracer supporting global illumination, PBR materials (principled BSDF), and HDRI environment lighting for publication-quality imagery. The framework excels at volume rendering of structured and unstructured grids via OpenVKL, isosurface extraction, and particle and molecular visualization. ISPC (Intel SPMD Program Compiler) enables vectorized execution that exploits SSE, AVX, and AVX-512 on modern CPUs, while MPI-based distributed rendering scales across hundreds of nodes for petascale data visualization.

OSPRay is deeply integrated into the scientific visualization ecosystem. It serves as an alternative high-quality rendering backend for VTK, ParaView, and VisIt, enabling users of those tools to produce ray-traced imagery with global illumination and physically accurate shadows without changing their workflow. Intel Open Image Denoise integration provides fast AI-driven denoising for the path tracer mode. The framework is actively maintained by Intel with regular releases and a growing user community in HPC, scientific visualization, and architectural visualization.

Quick Start

Download from https://www.ospray.org/downloads.html or build from source with CMake

Community & Resources

Performance Benchmarks

No benchmark data available for Intel OSPRay yet.

Benchmarks will be added as more renderers are tested across our standard scene suite.

Learn about our methodology