Rendering with Blender 3.6
Blender is a comprehensive open source software for 3D graphics, animation, rendering, post-production and interactive applications. It offers everything you need to create, edit and render complex 3D scenes. In addition to the classic 3D functions, Blender also includes tools for video editing, sculpting, UV mapping, texturing, rigging, particle simulations, physics and fluid simulations as well as basic game engine functionalities. Thanks to this enormous range of functions and its free availability, Blender is used by beginners, ambitious artists and professional studios alike throughout the media and design industry.
Before the benchmark results can be properly classified, it is worth taking a brief look at the underlying computing frameworks CUDA, OptiX, HIP and OneAPI. These technologies were developed for highly parallel calculations (High Performance Computing, HPC) and form the basis for GPU acceleration in the Cycles renderer. Blender 3.6 LTS uses different compute backends depending on the hardware used in order to optimally distribute the computing load between the CPU and GPU. Which backend is used depends on the graphics card and the installed drivers – and this has a significant impact on performance in the benchmark.
NVIDIA relies on CUDA as a proven basis and OptiX as a ray tracing extension. AMD uses HIP as the modern successor to OpenCL, while Intel offers complete GPU support in Blender for the first time with OneAPI. The respective degree of optimization of these interfaces is decisive for how efficiently the hardware works in the Cycles renderer. This is why NVIDIA usually achieves the highest frame rates in GPU rendering, AMD is in the solid midfield, and Intel is currently still heavily dependent on software updates to fully exploit the available potential.
CUDA is NVIDIA’s proprietary interface for GPGPU computing and has been the standard for general GPU calculations in Blender for many years. It is supported on almost all NVIDIA cards from the Kepler generation onwards and delivers stable, reproducible results. In Cycles, CUDA serves as a proven basis, especially when OptiX cannot be used or older GPUs are in use. OptiX, on the other hand, is NVIDIA’s specialized ray tracing API that accesses RTX GPUs and enables significantly accelerated path tracing thanks to its RT cores. OptiX offers massive performance advantages over CUDA, especially in scenes with a lot of light and shadow calculations.
AMD’s HIP (Heterogeneous Compute Interface for Portability) replaces the outdated OpenCL and forms the current backend for Radeon GPUs from the RDNA2 generation onwards. Although the implementation is not quite as mature as NVIDIA’s solutions, it shows noticeable progress with every Blender version. In the benchmark, this means that Radeon cards offer solid raw performance in HIP mode, but often lag somewhat behind the CUDA and OptiX results in practice.
Intel uses OneAPI, an open, cross-platform architecture for heterogeneous computing. In Blender, it serves as the primary backend for GPUs from the Arc series and allows the Xe cores to be used for path tracing tasks. Performance is currently still heavily dependent on driver maturity and integration in Blender, but improves with each version. For the Arc Pro B50 and B60, OneAPI is crucial to enable GPU-accelerated rendering at all – and at the same time ensures that Intel’s cards become increasingly competitive in Blender benchmarks.
In the Cycles renderer of Blender 3.6, NVIDIA uses CUDA and OptiX, AMD works with HIP, and Intel relies on OneAPI. These different compute backends explain the sometimes significant differences in performance between the workstation GPUs tested.
Conclusion
I have deliberately limited myself to a single benchmark for rendering, as this shows the limits of the tested cards. All three models belong to the entry-level class of professional workstation GPUs, which means that the pure raw performance is not sufficient for complex rendering scenes. Even more serious, however, is the limited memory: both the NVIDIA RTX A1000 and the AMD Radeon Pro W7500 only have 8 GB of VRAM. This memory reaches its limits at the latest with more complex meshes or high-resolution textures, which either leads to greatly reduced performance, rendering errors or even crashes.
The selected benchmark illustrates this problem very well, as it depicts typical rendering scenarios with extensive geometries and textures. The results are therefore representative of the entire class of these cards: the performance is sufficient for simple tasks, but they quickly reach their technical limits in professional workflows with higher model and texture complexity.
- 1 - Intro, overview and technical data
- 2 - Test system and equipment
- 3 - Teardown: PCB, topology and components
- 4 - Teardown: Cooler and fan
- 5 - Teardown: Material analysis and TIM testing
- 6 - Autodesk AutoCAD
- 7 - Autodesk Inventor Pro
- 8 - PTC Creo
- 9 - Dassault Systèmes Solidworks
- 10 - Autodesk Maya
- 11 - SPECviewperf 15 (2025)
- 12 - Adobe Photoshop 26.10
- 13 - Adobe After Effects 2025
- 14 - Adobe Premiere Pro 25.41
- 15 - AI benchmarks (AI Vision, Image, Text)
- 16 - Rendering
- 17 - Temperatures, clock rate, power consumption, noise
- 18 - Summary and conclusion
26 Antworten
Kommentar
Lade neue Kommentare
Urgestein
Veteran
1
Veteran
Urgestein
Urgestein
Urgestein
Urgestein
1
Mitglied
Urgestein
Urgestein
1
Urgestein
Veteran
Urgestein
Veteran
Mitglied
1
Alle Kommentare lesen unter igor´sLAB Community →