Cinebench is a free hardware benchmark from Maxon that measures how fast a computer’s processor and graphics card can render a 3D scene. It runs on Windows with Intel or AMD processors, Windows on Arm64, and macOS on both Intel and Apple silicon Macs. It gives owners, IT buyers, reviewers and hardware makers a repeatable score based on real rendering work.
A rendering workload instead of synthetic tests
Cinebench is a real-world benchmark built around common user tasks. It renders a scene built with Cinema 4D and Maxon’s Redshift engine, the default renderer in Cinema 4D. It uses the same rendering code that ships in Maxon’s own 3D software. Because the workload matches common Cinema 4D and Redshift tasks, the result also gives an indication of how the hardware will perform in those applications.
The scene in the multi-threaded CPU test is far heavier than in older Cinebench generations. This keeps modern many-core processors busy for a meaningful amount of time. Score values now use a new range, so users cannot compare results from the current Cinebench with scores from earlier releases.
Processor tests and what each one isolates
Cinebench splits CPU measurement into separate procedures, each answering a different question about the chip:
- Single thread renders on one thread, showing the speed of an individual core for tasks that cannot be split up.
- SMT single core runs the single-core procedure on a core with simultaneous multithreading enabled. This measures the gain SMT adds over plain single-threaded execution.
- Multiple threads spreads the render across the processor, reflecting performance in heavily parallel work such as final 3D renders.
Users can run the tests one at a time or all in sequence.
GPU benchmark with Redshift
The Cinebench GPU test renders the same scene with Redshift on the graphics card. Supported hardware includes NVIDIA cards through CUDA, including data center GPUs, and AMD Radeon cards on Windows. It also includes Apple silicon GPUs through Metal, with hardware ray tracing on newer Apple chips. The workload is built to occupy even machines with several high-end GPUs for minutes, so it runs slowly on entry-level graphics hardware. Since CPU and GPU workloads differ in demand, the GPU score is not meant to be compared against the CPU score from the same machine.
Sustained load and stress testing
By default, each Cinebench test runs for a minimum duration. This exposes whether a computer keeps its speed once it heats up. It also shows whether its cooling and power delivery hold under long periods of full CPU load. The Advanced benchmark option lets users set their own, longer minimum runtime, turning Cinebench into a stability and thermal stress test.
Results naturally shift with temperature and power state. A cool, freshly started machine usually scores higher than one that has been under load for hours in a warm room. For consistent numbers, Maxon advises closing other programs and pausing antivirus scans and disk indexing before a run. Small variations from background processes that cannot be stopped are normal.
Command-line runs
Users can also start Cinebench from the command line. Its options let users select a specific test, run every test in order or set the minimum duration. In this mode, Cinebench prints the score to the console and does not write it to a file. Users must therefore capture it from the console output.
What the score does not cover
Cinebench reports rendering performance only. It cannot read the power drawn from the wall, so it does not show how much energy a render consumed. Efficiency comparisons need separate power measurements alongside the Cinebench score.
