Cinebench Download — ComputerBase
Cinebench von Maxon ist der bekannteste Multi-Core-Benchmark für CPUs. Das Testprogramm basiert auf dem professionellen Rendering-Tool Cinema 4D. ComputerBase nutzt Cinebench seit Jahren als einen der Standard-Benchmarks in den CPU-Testparcours. Mit Cinebench R20 wurde 2019 eine vollständig überarbeitete Fassung veröffentlicht.
Das ist neu seit Cinebench R20
Seit Cinebench Release 20 wird eine deutlich größere und komplexere Test-Szene als noch in Cinebench R15 verwendet, die eine etwa achtmal so hohe Rechenleistung beim Rendern benötigt und vier Mal so viel Speicherplatz belegt. Mit Cinebench R23 wurde die Rendering-Engine deutlich angepasst. Messergebnisse aus R15, R20 und R23 sind deshalb nicht miteinander vergleichbar.
Die Mindestvoraussetzung ist eine CPU mit SSE3-Befehlssatz. Cinebench ist für Windows 7, 8 und Windows 10 sowie macOS ab Version 10.11.6 verfügbar. Das Betriebssystem muss in 64 Bit vorliegen. Seit Cinebench R23 wird auch Apple M1 unterstützt, Apples erster Notebook-Prozessor auf ARM-Basis.
Cinebench – Performing Render Test
Die Community von ComputerBase hat Cinebench R23 ausführlich getestet. Der Artikel Neuer CPU-Benchmark: Die Community testet Cinebench R23 enthält hunderte Testergebnisse ganz unterschiedlicher CPUs der letzten gut zehn Jahre. Auch für Cinebench R20 gibt es einen Community-Test.
Mehr Infos: Website
Weitere Downloads aus der Kategorie Benchmarks
-
4,7 Sterne
Prime95 ist ein idealer Stresstest für Mainboard, Prozessor und Arbeitsspeicher.
- Version v30.8 build 17
- Version v30.8 build 15
- +2 weitere
-
5,0 Sterne
FurMark ist ein kostenloses Tool, mit dem sich Grafikkarten auf ihre Stabilität überprüfen lassen.
- Version 1.32.1.0
-
4,3 Sterne
UL 3DMark ist der Referenz-Benchmark für Grafikkarten.
Diese Version wird seit 2013 kontinuierlich weiterentwickelt.- Version 2.25.8056 Deutsch
- Version Systeminfo 5.56
-
4,9 Sterne
CrystalDiskMark ist ein beliebtes Benchmark-Tool für Festplatten und SSDs.
- Version 8.0.4c Deutsch
- Version 7.0.0h Deutsch
- Version 6.0.2 Deutsch
-
5,0 Sterne
Der Download enthält alle Savegames, die Leser zum Benchmarken auf Bitte der Redaktion benötigen.
- Version The Witcher 3 Next Gen
- Version Cyberpunk 2077
- +4 weitere
-
4,4 Sterne
Mit CapFrameX lassen sich CPU- und GPU-Benchmarks aufnehmen und übersichtlich darstellen.
- Version 1.7.0 Deutsch
- Version 1.7.1 Beta Deutsch
Maxon Cinebench R20 | Intel DevMesh
Marc Potocnik
Düsseldorf, Nordrhein-Westfalen
2
0
-
0
Collaborators
Cinebench R20 is a free cross-platform test suite that benchmarks the speed of your computers CPU cores.
Its the standard benchmark on the market, used by people all over the globe. I created the test scene: An everyday living room in early spring; worn out, used and flooded with light. …learn more
Project status: Published/In Market
oneAPI,
Graphics and Media
Groups
Creators Artists
Intel Technologies
Intel CPU,
oneAPI
Links [3]
Overview / Usage
Cinebench is a cross-platform test suite that benchmarks the speed of your computers CPU. Cinebench measures multiple CPU cores and modern processor features. Its the standard Go-To benchmark suite on the market. And it’s free. Users, Bloggers, YouTubers and Tech-Nerds all ver the Globe use Cinebench as a reliable universal benchmark to measure CPU speeds. I created the livingroom test scene for Cinebench R20.
https://www.maxon.net/en-us/products/cinebench-r20-overview/
Methodology / Approach
After having created the OpenGL test scene for Maxon Cinebench R15 (the car chase) Maxon asked me to contribute the testscene for Cinebench R20. The scene should be highly realistic, very fast to render and without preparation time (e.g. for Global Illumination). So I focused on the living room scene of my animation project «Oberbilk» (https://vimeo.com/231428985): an everyday living room in early spring, it’s worn out, it´s used and flooded with light.
For Cinebench R20 I optimized this scene to be rendered as a still with Cinema 4D Standard Render. The finalized scene rendered in around 100 sec. in 1280×720 on a low-/mid-level machine — and now serves users around the world to test their computers. PS: You can see some eastereggs hidden in the scene in the stills under http://www.renderbaron.de/stills.html. 🙂
Originally the living room scene of «Oberbilk» was created for a tutorial for british „3D Artist“ magazine, #94 (2016).
Load the free PDF here and have a look behind the scenes of this project: renderbaron.de/neues/renderbaron_3DA_094.pdf
Technologies Used
The animation «Oberbilk» is created with Cinema 4D. The camera movement was tracked with SyntheEyes from a real life smartphone camera and then imported via Python into Cinema 4D. Lighting is done with direct illumination only, no Global Illumination was used. The scene was rendered in Cinema 4D with Physical Render (Intel Embree accelerated). Rendering hardware were two HP Z-840 and three HP Z-620 workstations equipped with fast dual Intel Xeon processors.
Other links
We are sorry, but without JavaScript we are currently unable to display the latest activity feed.
Please, enable Javascript in your browser.
Retesting Cinebench — downgrading AMD results and possible version spoofing
This material was written by a website visitor and has been rewarded.
The first testing showed that Cinebench with each version increased the advantage of Intel processors over AMD, the situation is similar to CPU-Z , and I decided to do more detailed testing.
I tried to take into account the comments on previous articles and make a test taking into account everything that comes out to be taken into account, but looking ahead I will say that nothing has changed dramatically in terms of the difference between the results…
recommendations
This time I will be using the i3-10105F and the R7 2700X with a fixed all-core frequency of 3.9 GHz, and the R7 2700X will be running with different core configurations.
I’ll also add R5 1600AF to the comparison purely for informational purposes, but it runs under Windows 7, because there will be no results in Cinebench version 23, since it is forbidden to work on systems lower than Windows 10.
AVX / AVX2 and libraries from Intel
First, a little about instructions, Cinebench uses AVX instructions, so all claims in the comments about weak AVX in AMD Ryzen (zen +) will be ignored, since Zen + has no problems with AVX, there are problems only with AVX2 which is not used.
Instructions that Cinebench R20 uses, you can immediately see that Cinebench R20 uses libraries from Intel (libmmd.dll):
0022
The first step is to download and unpack all versions of the test, and at this stage I encountered a substitution of the Cinebench 11.5 version, the developers changed the benchmark and it does not match the archived version that was on the Guru3D «mirror».
From the changes, at first glance, only the result in the database for the Xeon W5590, it became larger in the «new» version, and the W5590 developers tested on Windows XP instead of Windows 7…
At the same time, the developers forgot to change the assembly version for the Xeon result W5590 from the database.
I also note the copyright of 2009 in both assemblies, although the assembly from the «mirror» was compiled in 2009, and the assembly from the official archive in 2010.
CPU-Z and operating parameters
i3-10105F, R7 2700X, R7 2700X (4+0), R5 1600AF
All processors except R5 1600AF operate at 3.
9 GHz, in case of Intel this is setting the multiplier to 39 on all cores, in case of AMD this is the P-State setting.
HPET (High Precision Event Timer)
In the comments to the previous test, some people wrote that due to HPET there could be different results, I thought about it for a long time, and decided not to touch anything for a number of reasons:
1) some HPET can influence, and will use the default settings in 100% of cases.
2) I use the same SSD with the same operating system, and I do all the tests in one go without unnecessary reboots, etc. nine0004
3) The comparison is relative, even if HPET affects, it will affect all collected results within the run.
4) Taking into account the second and third points, the situation with different results is excluded when taking from different people seemingly with the same configurations and comparing getting a discrepancy.
I don’t see the point in doing HPET testing within the scope of this article.
Comparison of results between different versions
Compare results of different versions and compare difference between results of different versions this is not the same as .
Test results
First comes i3-10105F, then R7 2700X, R7 2700X with CCX disabled, R5 1600AF.
Cinebench 11.5 (RC23933)
Cinebench 11.5 (from the official site CB25720)
Cinebench 15 (RC8332)
Cinebench 20 (281795) 9000 9000 9000 9000 9000 9000 9000 9000 9000 9000 9000 9000 9000 9000 9000 9000 9000 9000 9000 9000 9000 9000
Cinebench 23 (330542)
Comparison of results was identical to i3-10105F, and the situation with the results remained identical to
as in the previous test …
there is no one stream of obvious underestimation of the results of AMD processors.
nine0004
In the mode of using all processor threads, there is a clear tendency to overestimate the result for the Intel processor and underestimate the result for the AMD processors.
I did not compare the old version of Cinebench 11.5 with the rest because of the large error.
Compared to the i3-10105F, the performance of AMD processors is obviously underestimated, but in single thread mode there is no obvious difference when the frequencies are fixed and Boost technologies do not interfere with work. nine0004
And if you look at the old version, you get the feeling that Cinebench developers are rushing between AMD and intel, but even in the old version, the relative results of AMD were better than in the new versions, although you should not forget about the error of the old version and Win7 in the case of R5 1600AF.
Results
Even after a more thorough retest with fixed processor frequencies in different configurations, the decline in multi-threaded performance remained just as obvious,0008 as in the previous testing , where the processors worked in stock Boost mode, the only thing that became ambiguous was the difference in single thread mode.
The fact that the developers of Cinebench version 20/23 used libraries from Intel is a fact, and it is quite likely due to this fact that AMD processors in these test versions went negative relative to Cinebench 15 when using all threads, while the processor from Intel went only in plus.
About AVX/AVX2, it’s is not the same thing, the first Ryzen had problems with AVX2, but they didn’t have any problems with AVX, and this is important since Cinebench does not use AVX2 AVX is trying to justify the underestimated results in new versions of Cinebench.
And yes, I tried to lower the AVX2 multiplier in the BIOS with the i3-10105F, and I did not find any difference in the results with the AVX2 multiplier lowered, which once again confirms what was said earlier. nine0004
I think it’s better to look towards open source benchmarks than towards Cinebench and CPU-Z which is closed source…
That’s all, thanks for your attention and comments from which new ideas for testing appear.
This material was written by a site visitor and has been rewarded.
How to run Cinebench to test your PC
We’re here at Edge use benchmark utilities to compare PCs against each other and find out which ones are the most powerful. But this is not their only goal. You can also run them just to see how your own PC performs against the competition and make sure it works as it should. nine0004
Cinebench R23 is one of the easiest utilities to use for beginners. This program uses your processor to render a 3D image and evaluates how well it does it. It’s free and only takes 10 minutes.
Here’s how to do it.
This is what Cinebench looks like in the app store.
How to download Cinebench
If you are using a Mac, search for Cinebench in the App Store. If you are using Windows, you can find it in the Microsoft Store. Users of any platform can also download it. from Maxon’s website directly.
nine0004
If you are running Windows 10 or 11, your system must have a 64-bit Intel or AMD processor with SSE3 support and at least 4 GB of RAM to run Cinebench. If you are a Mac user, you need to be running macOS 10.13.16 or higher; you also need a 640-bit processor and at least 4 GB of RAM.
Note: You may find different versions of Cinebench, but at the moment Cinebench R23 is the latest and best to use.
This is what it looks like during the test. nine0158
How to run Cinebench
Once you have downloaded the program, open it. You will see two Start buttons in the top left corner. Click next to CPU (multi-core) to check the multi-core capabilities of your system and CPU (single-core) to check its single-core capabilities. Both of these tests will take about 10 minutes.
You can also run tests up to 30 minutes long if you want to see if your system can maintain its performance for a longer period of time.
To set this up go to File > Extended test. A drop-down menu will appear on the left side labeled Minimum test duration . Select 30 minutes from this menu. Then run whatever test you want.
For the most accurate result, ensure that no other programs or background processes are running while Cinebench is running and that your device is connected to a network. Any settings on your computer (power profiles, fan settings, etc.) should be maxed out. nine0004
This is what it looks like when you receive your bill. I had a bunch of stuff open while my MacBook was running this, so this isn’t the most accurate estimate.
How to see your Cinebench result
Once your device completes the test (this may take a while, so be patient), a score will appear next to the start button you pressed earlier. In the bottom left corner, Cinebench will show you how your processor compares to a list of similar chips.