Author: Pieter-Jan Plaisier
Since 28th June 2007, the HWBOT database offers users the option to indicate what cooling has been used on both CPU and GPU. Many overclockers are using this feature, which is quite positive as it not only gives us valuable information on the overclockability of hardware in relation to the temperature (think: Athlon64 versus Cedar Mill), but can also give us a better view on the popularity of the hobby named ‘Extreme Overclocking’ â€¦
But how do we define ‘Extreme Overclocking’? Is it, as some say, just optimizing a given system as much as possible in an attempt to maximize the system’s performance regardless of the hardware or cooling method used, or is it as others claim applying extreme cooling methods to achieve the greatest frequency. Both points of views are debatable, but for this article, we have chosen to follow the latter group and consider extreme overclocking to be using subzero cooling methods to maximize the system’s operating frequency. To put it in terms of cooling methods, this varies from phase-change to liquid helium.
(Click to enlarge)
When we look at the number of overclockers who use subzero cooling, we notice an increase in 2008 and 2009, which is pretty much equal to the total amount of results submitted using subzero cooling. The figures of 2010 are, for obvious reasons, very incomplete as we only have data from the first three months. Although it’s incomplete we can see a growing trend: the amount of GPU and CPU/GPU subzero cooling overclockers and submissions are almost at the same level as 2009, with still 9 months to go, which could mean that 2010 will be the most extreme year we have seen thus far.
The absolute figures are possibly even more interesting. In 2007, there were roughly 550 extreme overclockers submitting scores to HWBOT, which increased to almost 1100 overclockers in 2009. This year, we already have 630 extreme overclockers having submit at least one submission using subzero cooling on the CPU. The number of complete CPU/GPU extreme overclockers has risen from 117 in 2007 to 207 in 2009 … and already 172 this year.
In terms of results, we can proudly say that in 2009 25,000 results were submitted using subzero cooling on the CPU. After 89 of the 365 days in 2010, we’re already at 10432 scores which means that if the current ratio is maintained, we’ll have almost 43,000 submissions with the CPU below 0Â°C. Amazing!
To have a better view on the ‘booming’ of extreme overclocking, we have performed another query trying to track the extreme overclocking behavior more in detail. We changed the cooling type from ‘subzero’ to ‘liquid nitrogen’ and measured the amount of results and different overclockers using the LN2 over a 2-month period from June ’07 to March ’10.
(Click to enlarge)
The amount of people switching to liquid nitrogen for cooling down the CPU has been rising quite rapidly since January ’09 after having decrease a few months earlier. When looking at the figures for the GPU LN2 overclockers, we notice that since January ’10 (introduction of rev3) quite a spectacular increase has had place as the amount of overclockers using LN2 for the GPU has almost been doubled!
The increase in results is also quite spectacular, especially for the CPU-LN2 submissions. A few paragraphs back, we estimated the total number of extreme cooling submissions for 2010 to be around 43k, but taking the exponential (describing form of curve, not the math) growth into account, this may actually be a lot more. Also noteworthy, in a previous article named “Is LN2 on GPU really worth the trouble?” one of the assumptions was that however was using LN2 on the GPU would also be using some form of extreme cooling on the CPU. In all the four charts of this article, we can see that this assumption is not that far off from the truth given that the figures of GPU-LN2 and CPU/GPU-LN2 are almost the same.
After having seen the ‘proof’ for this assumption, maybe we can now state that the process of becoming an extreme overclocker involves using liquid nitrogen on the CPU before using liquid nitrogen on the GPU. Or, to make the statement a bit more bold: extreme 3D overclocking is more difficult than extreme 2D overclocking.