Vadim Chobanu

I was an advocate of watercooling and always believed in it. I believed in it so much that I even invented and developed my own waterblock for the VGAs. If I was to have that product 3-4 years earlier, probably it would have been successful, now I struggle to see a long term demand and wont bother investing any more time or efforts or even think about it.

Probably many will say that I am gone mad, yet I have valid reasons for that and below I will give the technical explanations behind it.

1) The CPU manufacturing process is shrinking and the current generation of Intel CPUs is at 45nm, due to move to 32nm later on this year. I hear many say ’so what has the manufacturing process to do with the actual watercooling being obsolete?’ It’s simple – more components packed together in the same size will need quicker heat exchange by the exactly same area of contact and the copper commonly utilised in the industry is starting to bottleneck and no matter how quickly the liquid will be pumped, through the CPU block and how cold will that be, the actual CPU temperature will be quite different from the overall ambient liquid temperature.

Example: back in 2006 when Core 2 Duo (65nm launched, the watercooling made a big difference to how much you could push your system to. To stably overclock the first batches of E6600 CPUs to 3.6Ghz on a commercially available air cooler was almost impossible, yet possible with watercooling. It actually made a big difference to the overall CPU temperatures. Now, with the launch of the Core i7 the watercooling struggles to give you as much quantitative benefit when compared to aircooling and I believe that will shrink even more when the manufacturing will move to 32nm. Unless of course diamond or diamond based heat exchangers and compounds will be widely accepted as a cooling standard. One of the latest attempts in this domain was reported by Custom PC not long ago - http://www.custompc.co.uk/news/605550/researchers-create-new-thermal-material.html

If it will be cost efficient, then there is still hope for watercooling to be effective for a few more years to cover the 22nm CPU manufacturing, but probably not as much for the 16nm later on.

Fact: If you look below at the E6600 world records you will see all attempts were made using one sort or another of extreme cooling, such as liquid nitrogen or cooling cascades.

Now if we do the same search for the newer core i7 CPUs, we would notice that in the top searches we get even aircooling and watercooling alongside the liquid nitrogen. That makes a watercooled Intel Core i7 CPU 27Mhz faster than watercooling, leaving even some phase change cooling results behind.

A ‘better’ alternative to watercooling is considered Peltiers/TEC (Thermo Electric Cooling) which is being adapted for use in PC systems by the Canadian Cool IT and in the UK an interesting attempt at using the concept was done by Kobalt Computers, where a Core i7 CPU was overclocked to 4.4Ghz with fairly interesting temperatures as reported by media. I was never a believer in TEC due to the efficiency limitations (the best peltier will only achieve 67% conversion efficiency and that is behind other technologies (PSUs are currently at 80-90%), yet the concept works. Maybe its not the quetest cooling, but it works.

Achieving 4.4Ghz from a Core i7 with good temperatures is pretty impressive, yet the price or availability for the full system was not released and I would imagine they are quite prohibitive in the today’s climate when no-one wants to spend money. A smartly built aircooled system with the same components wont be that far of from those results and that is fact.

My second argument today against the need for extreme speeds is also simple – what for?

A few years back it was understandable – to play the latest games on the highest settings you needed to overclock heavily and always have tomorrow’s technology. Today there are not that many games that wont play very well on a aircooled and overclocked to 3.6Ghz Core i7 with a few high end VGAs from AMD or Nvidia in it.  So the only purpose of watercooling would remain  - noise removal, which will win against the TEC concept.

I understand that not everyone will agree with me, please feel free to comment