Search

PSA - Intel 12th Gen Overclocking

We recently covered the Intel 12 Gen release and it is clearly competitive with AMDs best offering, in many cases even better. Intel CPU’s historically have had a great overclocking potential to push way past its stock performance, so we set out on a journey to test it and see if it’s worth overclocking.



In the interest of providing the best value to you and saving your time, we are not going to beat around the bushes and tell you as it is - should you overclock 12th Gen Intel CPUs - in general, no. But there is performance to be gained so if you are interested to get the insight into overclocking potential stay with me as we go over it.


Let me give you a brief into our system set-up. We are using a top of the line Intel i9 12900K CPU, running on ASUS Maximus Z690 Hero motherboard with Geil DDR5 memory. We will be running this memory at different speeds, and show you the overclocking settings shortly. Both cooler and GPU are the same across all systems. As far as AMD and last gen Intel systems are concerned, those we covered in our main CPU review (you can check it out in this link).


In article video we will be using 5 different setting configurations for the 12900k. Stock, then two simple memory overclocks: 4800 and 5200 mega transfers respectively. And finishing up with two memory and CPU overclocks: one with auto OC within Intel extreme tuning utility which boosts CPU clock speed by 100 MHz and the other being manual OC.

With 12 Gen Intel has implemented hybrid core design; now there are performance cores called P-cores and efficiency cores called E-cores. This makes overclocking slightly different from usual, but with the help from Intel’s overclocking utility it is pretty quick and can tweak all the settings very nicely. For our complete manual overclock we have gone with P cores at 5.2GHz, E cores at 4.1 GHz and voltage offset of -0.05.

As you may see - we didn’t really focus on efficiency here, rather just trying to squeeze every bit of performance out of this chip. With this in mind let’s check out the results, starting with some gaming benchmarks. First up is Horizon Zero Dawn and here we see that the spread between all of these set-ups is pretty small. From stock to leading OC there is only about 5% difference.


Next up is Shadow of the Tomb Raider and here Intel 12900K with stock settings has lower average FPS than AMD counterpart. Simply overclocking the memory improves performance by about 6% and we see better improvement from faster memory and CPU overclocking. The total improvement here is up to 12% which is pretty good. Take note of that for later.


Lastly we have Doom Eternal and here just like in Horizon Zero Dawn there is only 5% improvement from stock to manual overclocking, but that is not the whole story on all of the games. If we switch over to power efficiency across these games, the results are very different.

In our main review of the next gen CPUs from Intel we found that they are really power efficient when it comes to gaming as compared to both AMD and last generation Intel CPUs.

Naturally overclocking changes this, with exception to RAM overclocking, as that has a reasonably low impact on power consumption, but it can have a great impact on performance. For example, in Doom Eternal from overclocking the RAM we gain between 1 and 4 percent improvement on FPS per 10 watts, but the moment we overclock the CPU - our efficiency goes down by over 15% as compared to stock.


In Horizon Zero Dawn overclocking RAM netted no difference in performance but overclocking the CPU reduced power efficiency by 18% which is equivalent to shooting yourself in the foot. I would probably avoid it.


In Shadow of the Tomb Raider, if you remember from earlier, there was a considerable improvement from both RAM and CPU overclocking and it shows on the power efficiency too. At the top of the table we see RAM OC at 4800 MT with 29% improvement on FPS per 10 Watts. The moment we overclock the CPU our efficiency drops from stock by 12 to 14 percent, but if we compare between manual CPU overclock and memory overclock then the difference is a whopping 50%.

Not sure if it's worth doing at all, but let's move on to productivity.


And if you are finding our work useful and want to support us, please consider subscribing.


The first benchmark is Cinebench R23, and here we find the best single core score is with memory OC, but the difference between results is very minimal. When it comes to multi core performance this is naturally best with both types of overclock, even though it's just a few percent improvement.


Next up is Vray and here AMD still has a solid lead with fully overclocked Intel chip lagging behind. Just like in most of our tests the difference between all of these settings is very minimal.


Last test we have here is Blender and it basically gives the same old results. Slight improvement from manual OC on both RAM and CPU but nothing too crazy. We also analysed the data further while running the longer Barcelona test and found that while auto OC is enabled, the CPU is always staying above 90 degrees Celsius in the room with an ambient temperature of 28 degrees.


However, with manual overclock it immediately hits 100 degrees and stays there, you can see little dips, this is where CPU is actually thermal throttling and clocking itself down.


If we look at the power efficiency, the story is even more disappointing. Under full load the manually overclocked 12900k is pulling 270 Watts at the beginning before hitting thermal limits, then dropping down to low 260s. It is exactly double of what AMD’s Ryzen 5950X is pulling so there is clearly a long way to go before these chips become competitive in productivity.


Which brings us well to the statement that I mentioned at the beginning of this video - why do I think it is not worth overclocking this CPU? It is clear Intel has already optimised its performance and built that into their boosting algorithms. There is maybe 5% more performance to push at the expense of thermals and power, however I’d say is not worth it for almost anyone - it is fun to go out and set some high scores in Cinebench against AMD fanboys but that is about it.

If you are looking for more performance then I would recommend looking at getting the fastest possible memory with lowest latency, as this will have the most impact on performance without heating up your house for no reason.

_______________________________

This post contains affiliate links. An affiliate link means we may earn advertising/referral fees if you make a purchase through our link, without any extra cost to you. Thank you for your support!

0 comments