Cinebench Good Scores

[Rafa Ostermann]

I got my CPU with my new Mobo a few weeks ago, everything is working fine and i have stellar performance in games ( atleast i think so ), i decided to run a cinebench R23 and 3D mark to see if my new cooler Z73 kraken is doing his office, and he is temps are great !

I found some tests on google about the 11900k and cinebenchR23 they are around 16209, with just ABT on, mine with same option is around 15602 but that's fine, the thing that is worrying me is a friend of mine have the same CPU as me and he also have ABT on, and his CPU score on 3dmark is about 14000, mine is only 12500 why is there such difference ?

Those tests don't tell in real life how the CPU performance is with real workload... There would be always differences in those performance tests... even for equal components... many other factors may alter the results... ABT (Turbo and other technologies) is not guarantee always equal... Mobo bios settings and power are also relevant...

I have some systems with different CPUs... For one of them, my nephew and I have the same CPU, same psu, same cooler, same case... and he gets lower temps, but I get better results in performance tests (and higher temps)... both are happy with our CPUs... having different results doesn't matter much if you are ok with the work the system does...

You are comparing different tools for tests... in the first one I see you are fine with the result.... but not with the second tool... if you test more and more test I guess you would have many many different results ranging from very good to very bad... For me, it doesn't worth to spend time in such as anguish...

And yes i checked my Silicon quality ( SP in asus bios ) and i'm only 50, i have the worst score possible which explain why when i did run C23 and 3dmark my CPU points were lower than average, but tbh as your post confirmed what i understood by myself it doesn't matter ...

To be honest i'm new to the whole CPU and benchmark thing i'm learning everyday, i'm just not lucky on my SP quality, my buddy prolly have 90 SP points or something to score that high on synthetic benchmark.

Intel does not verify all solutions, including but not limited to any file transfers that may appear in this community. Accordingly, Intel disclaims all express and implied warranties, including without limitation, the implied warranties of merchantability, fitness for a particular purpose, and non-infringement, as well as any warranty arising from course of performance, course of dealing, or usage in trade.

To get started, you will need to navigate to the BIOS. Unfortunately, now you cannot use Ryzen Master to do this, but AMD claims that this will be part of Ryzen Master in their future releases. In the PBO section, you will need to adjust some settings.

Max CPU Boost Clock Override should be set to 200MHZ. This allows for higher clock speeds on single threaded workloads. My 5900x can hit 5.15 GHz with this setting on a single core. 5.15 GHZ is not a one-off number. I regularly see this during light workloads

Firstly, I started with negative 20 on all cores. This resulted in awesome Cinebench R20 scores but poor stability. I have then went to negative 15 on all cores. This was not bad, but I was experiencing a crash every now and then, especially when the PC is running cold and is able to push more clocks. It would run all day, but on boot, pushing it will instantly result a crash. This tells me that the algorithm was trying to push for more clocks, but the undervolting was too aggressive.

I then went to negative 10 on all cores and it is fully stable. Finally, I pushed negative 15 for those cores which are not first or second. This remained stable, and eventually I started changes the values slightly everday. Sometimes I go too much and get a WHEA BSOD (especially when the PC is cool and under light workloads).

Thanks for the article! Found it extremely similar to what I have currently been conducting with a Ryzen 9 5950X. Although I found that in the scaler menu one thing that confused me was in bios your cores start at 0. In Ryzen Master they are numbered 1-16. So in bios core 0 is actually core 1 in ryzen master. This will cause a mess if trying to push your best cores for single thread higher then others as I made this mistake in bios. And I received best results with my best cores set to +5 and all other set to -10. This did indeed boost my scores on cinebench and 3DMark. Just wanted to share the information.

Hi, thank you for this article! Great work. And finally, somebody who tried PBO 2 settings and shared his experiences with others ? Based on your article I tried both (the same negative for all cores, then different for some cores). But after a few weeks, I turned PBO 2 off. Yes, there was some performance gain. But at the cost of higher temperatures. Running BOINC at 100 % on all cores was 84C with PBO 2 vs. 74C without PBO 2.
So without PBO 2, my CPU can run at 100 %, and I barely hear the fan (Noctua NH-U12A). With PBO 2 turned on, my PC was noticeably noisier. Not as much as my previous PC, Noctua cooler is great, but the performance increase was not worth it.

Do you use the same EDC and TDC maximums for both CPU and SOC?
I have the same motherboard and CPU so just starting to play with the PBO settings.
Currently hitting 4.5Ghz all core with a max temp of 76C.

This was extremely helpful, thank you Albert! Following your PBO limits I went from 8200 to 8800-8900 C20 score on 5900x / X570 Taichi @ 4.5ghz with Dark Rock Pro 4 at 80C peak. -7 on stars and -5 second best cores on both CCD0 & CCD1 with -15 for the rest. This actually lowered my idle temp around 7 degrees after lowering my chipfan curve due to noise. This was a fun exercise for a casual overclocker. One question, how are you deducing how far to tweak each core, do you crash then dial the value back for each core? For your CCD1 do you still have a smaller offset for star & circle core?

Hey, first very good guide and tips on pbo2. Second I would like to ask you have you noticed any gaming performance gains using this overclocks? I mean ok benchmarks and multicore are one thing, but it would be interesting if somebody shown how to tweak 5900x or such to maximize gaming performance and fps. Thanks

Thanks for the quick reply. The pc sits at 35-37c on idle. The PC does have spikes up to 42-52 time to time when running daily tasks but stays pretty cool. The high temps are only a concern when gaming / running cinebench 23 testing.

I can write up a better explanation on how the curve optimizer works and why, if you need. In any case, you should know that the settings you have in your example are incorrect. Setting the curve optimizer properly is impossible the way you described.

Hi! Thanks for your feedback. This blog post merely documents what works for me, especially since this is original research rather than just rewording other guides, in the very early days of curve optimizer. In fact, I mostly still run the same settings, as they work for me.

Hi Albert,
I am very disappointed with my R20 multi core scores. I am only getting a score of 8400-8500 on 77c. I am using the 185 PPT, 125 TDC, and 170 TDC settings. For the curve optimizer, I am using 12 on the first preferred core, 16 on the second preferred core, and 18 on the rest. Can you help me?

Hi Albert,
I am very disappointed with my R20 multi core scores. I am only getting a score of 8400-8500 on 65-70c. I am using the 175 PPT, 125 TDC, and 150 TDC settings. For the curve optimizer, I am using 12 on the first preferred core, 16 on the second preferred core, and 18 on the rest. Can you help me? I closed background apps and I am using a 280mm Corsair AIO.

Thanks for sharing this. Im currently trying something with my 5900x based on your results (I have the asus x570i motherboard). I first set PBO limits to motherboard and then looked in Ryzen master just to see what they get set to.. I was suprised to see a wopping 1000w PPT by default!

So, I found out from the various internet sources that somewhat consensus is now to use CoreCycler ( ) as one of the test applications to ensure stability for Zen 3 processors. I ended up with curve optimization as per follows:

For some context, check out the AMD Threadripper 2990WX, a 32-core workstation processor designed specifically for highly threaded workloads with a score of 30,054. The 13900k bests it by over 35%! Here are some more results for comparison:

Time Spy Extreme scores can vary widely depending on the clock and cooler, but 13,000 with a single 3090 Ti is generally considered a good score, with the record coming from an overclocked liquid cooled 3090 Ti at 14,611. Suffice to say, this +33% increase over the previous world record is nothing short of astonishing, demonstrating an incredible generational leap for the RTX 4090.

Cinebench is a real-world cross-platform test suite that evaluates your computer's hardware capabilities. Cinebench Release 20 provides a more accurate measurement of Cinema 4D's ability to take advantage of multiple CPU cores and modern processor features available to the average user.

For over a decade, Cinebench has been a popular tool to evaluate hardware performance. Cinebench scores are used by system administrators to help make purchase decisions, journalists to review hardware, hardware manufacturers to optimize their latest products, as well as computer owners to evaluate their individual system.

Improvements to Cinebench reflect the overall advancements to CPU and rendering technology in recent years, providing a more accurate measurement of Cinema 4D's ability to take advantage of multiple CPU cores and modern processor features available to the average user.

The test scenario uses all of your system's processing power to render a photorealistic 3D scene (from the viral "No Keyframes" animation by AixSponza). This scene makes use of various algorithms to stress all available processor cores.

In fact, Cinebench can measure systems with up to 256 processor threads. This test scene contains approximately 2,000 objects which in turn contain more than 300,000 polygons in total, and uses sharp and blurred reflections, area lights, shadows, procedural shaders, antialiasing, and much more. The result is displayed in points (pts). The higher the number, the faster your processor.

c80f0f1006