CINEBENCH - HOW WE MEASURE

Cinebench is a real-world performance benchmark designed to reflect common professional 3D workloads. It uses a production-quality render scene comparable to what a 3D artist would create in everyday work. The scene has been carefully selected to represent a broad range of rendering tasks, including ray-tracing performance, shader execution, geometry processing, lighting and global illumination calculations as well as general scene complexity.

The workload is identical across platforms and devices, ensuring consistent and reproducible results. 

Cinebench 2026 results are calibrated to a new scoring range and are not comparable to results from earlier Cinebench versions.

This test executes one render thread per available CPU core, including logical cores created through Simultaneous Multithreading (SMT). This reflects a common usage scenario in Cinema 4D, Redshift CPU, Cinebench and other applications that are well optimized for parallel execution.

In this test, all CPU cores are fully utilized over an extended period of time. As a result, the processor and its cooling solution are heavily stressed. In addition to the raw CPU performance, this test can provide insights into sustained performance behavior, thermal limits, and the effectiveness and acoustic characteristics of the system’s cooling solution.

The Single Core test compared to the Single Thread test demonstrates the performance uplift of the SMT capabilities of the core. The Single Core test result uses the same score calculation as the Single Thread test.

SMT (Simultaneous Multithreading) is a technique that allows one physical core to appear as two (and in rare enterprise-class implementations, four) logical cores to the operating system. This is achieved by sharing execution units within the physical core while duplicating only selected resources, avoiding the need to replicate the entire core.

Performance gains from SMT depend on how efficiently software can utilize the shared resources. Early desktop implementations, such as Intel’s Hyper-Threading, typically delivered gains of around 15%. Modern CPUs can exceed 30% performance improvements in well-optimized workloads that avoid saturating shared execution units.

If no SMT is available, this test is greyed out.

The CPU Single Thread test measures the performance of a single execution thread running on a CPU core.

Strong single-threaded performance is critical for tasks that must run sequentially and cannot be parallelized. It also has a significant impact on perceived system responsiveness, often described as “snappiness”, particularly in user interface interactions, application startup times, and web browsing.

The GPU test executes the same Cinebench render scene on the graphics processor instead of the CPU. Depending on the architecture and capabilities of the installed GPU, this can result in significantly higher performance compared to CPU-based rendering.

This test highlights the GPU’s compute throughput, memory bandwidth, and ray-tracing capabilities (where supported), providing insight into how well a system performs in GPU-accelerated rendering workflows.

If no GPU is available, this test is greyed out.

Modern operating systems run hundreds of background threads in parallel. Some of these tasks can consume significant CPU resources when active, including virus scanners, enterprise security software, update services, search indexing, and web browsers.

For consistent and comparable results, check the Activity Monitor or Task Manager before running Cinebench and ensure that no background processes are placing a high load on the CPU or GPU, as this can skew benchmark results.