AllInfo

New CPU benchmark: The community tests processors in Cinebench R23

tl; dr: How fast is which processor in the brand new CPU benchmark Cinebench R23? The community is called upon to test models from AMD and Intel – as well as Apple Silicon in the future. The goal: A benchmark comparison with old and new CPUs and APUs from AMD and Intel as well as Systems-on-a-Chip from Apple.

Table of contents

  1. 1 This is new in Cinebench R23
    1. New features and still up to 256 threads
  2. The community can test with
  3. The results

Update 08/19/2021 10 : 51 o'clock

The article was about benchmark results of the CPUs tested by ComputerBase since April, including Core i7-11700 (test), Core i7-11370H (test), Ryzen 7 5700G and Ryzen 5 5600G (test), Ryzen 9 5900HX and Ryzen 9 5980HS as well as Ryzen 7 5800H (test), added.

Update 04/19/2021 5:15 p.m.

The article has been expanded to include the benchmark results of the new Intel Rocket Lake CPUs Core i9-11900K and Core i5-11600K (test) as well as Core i5-11400F (test) from the editorial team. In addition, further results from the community were taken into account, primarily from CPUs of the Intel Xeon type and older models such as AMD Opteron and AMD FX.

Update 01.03.2021 10:02 am

ComputerBase has expanded the overview with the first community benchmark results of the Intel Core i7-11700K sold by Mindfactory before the market launch. The results determined by the ComputerBase reader DriveByFM show the 8-core CPU in the multi-core scenario on the level of the 8-core competitor Ryzen 7 5800X, in the single-core run it is on the level of the mobile Tiger Lake Processor's Core i7-1165G7, which basically has the same architecture: Rocket Lake-S is based on the Sunny Cove cores of the Ice Lake architecture, Tiger Lake on the Willow Cove cores, which are only slightly adapted in terms of cache. The IPC is usually the same. As an iGPU, Rocket Lake-S uses the Xe architecture like Tiger Lake.

However, this value still seems too low, because the mobile CPU clocks with a maximum of 4.7 GHz, with the Core i7-11700K it should be 4.9 or even 5.0 GHz. From OEM circles, however, it can also be heard that the final microcode for the CPU has not yet been included in the current BIOS variants and that something will still change in terms of performance thanks to optimized turbo clock rates. That should also affect the multi-core performance.

This is new to Cinebench R23

Like Cinebench R20 (community benchmark) and older versions, Cinebench R23 (download) is also based on the technology of the well-known Cinema 4D rendering tool, but now on the basis of Release 23 among other things also the instruction set AVX and for the first time also the M1 processor from Apple supported. The minimum requirement, however, is still the SSE3 instruction set.

New features and still up to 256 threads

There is again a test run for the single-thread and multi-thread performance, both of which can now be selected individually. In Cinebench R23, Maxon sticks to the focus on scaling with many threads: Furthermore, a total of 256 threads are supported.

In addition, the benchmark's feature set has expanded and both The code and the compiler have been adapted, which is why, according to the manufacturer, older results should not be used for comparison.

The community can test with

How fast are processors in Cinebench R23? The editorial team is currently simply lacking free capacity to create a broad overview, so the Cinebench R20 (community benchmark) allows the community to show what it can do this time and to diligently deliver its own results. The multi-core and single-core tests must be carried out separately.

Note: The option “Advanced Benchmark” should be activated under “File”, the option “Minimum Test Duration”, which can be used for the new stability test of Cinebench R23, should be set to “off”.

Both benchmarks are immediately visible, but must be done separately be selected

Results are submitted using the form below and, after being checked by the editorial staff, are included in the diagrams. When submitting results, additional information such as the processor used and whether it has been tested in the factory state or overclocked (specifying the multi-core turbo) or, for example, in an eco mode is important. Pay attention to the correct spelling of “Core i9-10900K” or “Ryzen 5 5600X” – this makes it much easier to adopt valid results.

Note: With mobile CPUs and The selected TDP level should also be transmitted to APUs.

Examples:

There is no need to specify other parameters such as RAM, he has never been important in Cinebench.

Submit benchmark results Note: To submit benchmark results, please log in!

How does it work here? You publish your results as a forum post. In order to make our work easier, the data intended for the diagram must have a certain format. For this purpose, this form creates a forum post, which you send with additional screenshots and details (please do not change the generated part!).

1. Your system

Use the same formatting like already existing diagram entries (and note the information in the text if necessary). You can write details in free form in your forum post.

2. Your measured values ​​(without unit)

I have my entries checked again for correctness

The results

The following diagrams contain the first benchmarks of the editors (bold) and will be successively expanded to include community benchmarks. Systems with a CPU from Intel are marked in blue, systems with a CPU from AMD in red.

What has already been shown: The new engine can handle more cores even better. A Ryzen 9 5950X with 16 cores, for example, increases by 31 percent compared to the Ryzen 9 5900X with 12 cores, in Cinebench R20 it is 22 percent – and the gaps between other CPUs in the multi-core run are larger.

Cinebench R23 – Multi-Core Unit: Points

Cinebench R23 – Single Core Unit: Points

Was this article interesting, helpful or both? The editorial team is happy about it any support from ComputerBase Pro and deactivated ad blockers. More about ads on ComputerBase.

Downloads

Exit mobile version