The Testing Behind the Underclocking Experiment

Before diving into the results of the experiment, it's worth noting that the tests were conducted out of curiosity. There was no particular reason behind it apart from the fact that I wanted to see what would happen when severely underclocking a 5820K. It's not recommended to do this as it wouldn't make much sense and wouldn't lead to any significant benefits, such as power consumption decline or a cooler system. The CPU frequency drop to 1.2 GHz, for instance, is not beneficial and should be avoided.

The Experiment Setup

To conduct the experiment, I set up two scenarios: one where the 5820K was running at 1.2 GHz with 0.8 volts, and another where it was running at 4.6 GHz with 1.32 volts. Both systems were equipped with a minimal graphics card bottleneck, which should not have affected the results significantly.

The First Synthetic Test

I started by running the first synthetic test, 264, to check temperatures and ensure that both clocks were stable during the underclocked run. The underclocked system ran at 1.2 GHz on the left, while the overclocked system ran at 4.6 GHz on the right. It was found that the temperatures of the underclocked system remained stable throughout the test, with the CPU running at a temperature of X99.

The Overclocked System

On the other hand, the overclocked system yielded extremely uncomfortable temperatures, even though the voltage applied was perfectly fine and stable. The 240mm AIO used in the system couldn't dissipate heat quickly enough, despite its fans being at maximum RPMs. It's worth noting that I ran the CPU at 1.31 volts and 4.5 GHz for this particular test.

The Second Synthetic Test

I then moved on to running Cinebench R15, a synthetic test that measures CPU performance. The underclocked system resulted in a huge CV score cut of almost 75%, which is significantly lower than the overclocked system. This indicates that the underclocked system was not performing as well as expected.

The GPU-Bound Test

I also ran a GPU-bound test, Witcher 3 on ultra settings, to see how the CPU would perform in this scenario. The results were disappointing, with both systems struggling to maintain frame rates above 19 FPS on average. However, it's worth noting that even GPU-bound titles can be CPU-bound in certain situations.

The Final Test

Lastly, I ran a side-by-side comparison of Battlefield 1, a game that is known for demanding high-performance hardware. The results were somewhat unexpected, with the underclocked system performing slightly better than expected. However, it's still clear that the lower frequency was not as pivotal in this case, and the overall frame rate was still lower.

Conclusion

In conclusion, the experiment showed that severely underclocking a 5820K does not lead to any significant benefits. The temperatures of the underclocked system remained stable, but the performance suffered significantly. It's also worth noting that even GPU-bound titles can be CPU-bound in certain situations. If you're considering buying a locked Xeon or limiting yourself to locked SKUs, it's absolutely not worth it.

Don't Buy into the Hype

The most important takeaway from this experiment is that there's no point in doing what I did. The results are not surprising, and they only serve as a reminder that CPU performance is not directly related to frequency. If you're looking for a high-performance system, it's better to focus on buying a graphics card that can handle the workload rather than trying to limit yourself to locked SKUs or underclocking your CPU.

The Future of Gaming

As for the future of gaming, there are signs that things might be changing with the advent of DirectX 12 and Vulcan. These technologies are designed to accelerate performance and improve multitasking capabilities. However, for now, they remain a topic of discussion among gamers and hardware enthusiasts.

Final Thoughts

And that's it for this video. If you found this experiment interesting or informative, please give it a thumbs up. If you feel the complete opposite, I'd love to hear from you as well. Don't forget to subscribe to my channel for more content like this.

"WEBVTTKind: captionsLanguage: enbefore I say anything else I would like to point out that these tests were conducted because I  was curious that's it there's nothing else to it there's no rhyme or reason behind it other than  the fact that I wanted to see what would happen when I severely under clocked my 5820k I don't  recommend doing this it wouldn't make much sense apart from power consumption decline a much cooler  system and a much quieter one as a result there is absolutely no reason why you should drop your  CPU frequency down to 1.2 gigahertz or thereabouts doesn't matter what processor you're sporting what  CPU cooler you're using just don't do it why did I do it you asked whoa whoa wait wait yeah  that did it yeah magnets Oh what you're about to see is a side-by-side comparison between a 5820k  running at 1.2 gigahertz and point 8 volts that was the lowest I could drop the voltage without  the computer refusing to boot up and the same cpu running at 4.6 gigahertz and 1.32 volts with  a minimal graphics card bottleneck how large of a framerate disparity will we see in modern  triple-a titles let's start off first with a few synthetics I ran I 264 first to check temps and  make sure both clocks were stable on the left is the under clocked run the right the overclocked  one our temps literally did not change during the extent of our under clocked run something  I suppose shouldn't be too surprising given the voltage applied but it's still surreal to see an  X 99 CPU running at this temperature under load on the other hand the overclock scenario yielded  temperatures that were extremely uncomfortable the voltage applied was perfectly fine and stable but  the 240 millimeter AIO I was using just couldn't dissipate heat quick enough even with its pomp and  fans at their max RPMs for the record I run the CPU at 1.31 volts and 4.5 gigahertz for this very  reason next up with Cinebench r15 and as you can probably tell the under clock test resulted in a  huge cv score cut almost 75% by the time the tests finally finished we managed a mere 317 something  even my Q 6600 laughs at Geekbench mirror these findings roughly 70% cut for the 1.2 gigahertz run  in this case something worth mentioning here is how drastically the multi-core or change compared  to that of this single-core when you have more overclockable cores at your disposal overall  performance can significantly jump it's what makes the x99 chipset so appealing but only if the games  and programs being used are able to take advantage of the core surplus now let's throw a few games is  severely under clocked beast is it actually is it even a beast if it's running at 1.2 gigahertz the  answer is a big fat no check out this side-by-side comparison of the GTA 5 benchmark there is an  enormous framerate disparity and we can analyze it in two different ways the first is from a  simple minimum and average framerate bar graph 135 versus 48 frames per second on average is  a stellar representation of how important CPU frequency is to a degree there's another way we  can look at this frame times instead of looking at this in terms of frames per second let's try  milliseconds per frame in this case we're looking for a flat and consistent line with minimal spikes  the 1.2 gigahertz run was all over the place and finishes quicker because fewer frames were drawn  overall the 4.6 gigahertz run is a very flat and linear line devoid of spikes this is exactly what  we want to see a good indicator of a well balanced system for this game I'm running two gtx 1070 s  and sli by the way so GP bottlenecking should be minimal the only game on our list that's not  properly sli optimized is city skylines explained and verified right here but even with only one GPU  breaking a sweat it was a CPU that held us back substantially both the average and minimum frame  rates were utterly disappointing for the 1.2 gigahertz chip 19 FPS on average wouldn't even  cut it in the console sphere hue and check out that minimum seven single digits let's move on  how about a GPU bound title in the video I pointed to a bit earlier we verify that Witcher 3 on ultra  settings is a graphics card hog maxing out both 10 70s consistently in this scenario however the  CPUs abysmal frequency played far too greater role in frame rate reduction these disparities  aren't the worst we've seen thus far but it's important to note that even GPU bound titles  can be CPU bound in the right situations lastly we have battlefield 1 not a title I expected to  perform as well as it did frankly you're looking at a side-by-side comparison here can you tell  which is which in this instance battlefields craving Coors overclocks something I expect  will become the norm by the turn of the decade directx12 in Vulcan will accelerate this trend  but for now battlefield one remains one of the few games demanding more than four cores in which  case the lower frequency isn't as pivotal it's still clearly lower the overall frame rate but  not by the degree I expected it to to keep this one short and sweet don't do what I did there  is absolutely no point don't buy a locked Xeon running at one point whatever gigahertz and don't  limit yourself to locked SKUs in general if you're certain your graphics card can pull its own weight  if you thought this video was interesting give it a thumbs up give it a thumbs down if you feel the  complete opposite or if you hate everything about life be sure to click that subscribe button if you  haven't already I'll catch you in the next video this is Salazar studio thanks for learning with us\n"