interview with innovatek: watercooling reached it limits?

[DX2]

http://translate.google.com/translat...hl=en&ie=UTF-8

some interesting questions and one thats pretty close to most people here. Has watercooling exhausted is usefull potential? have a read and find out what innovatek says about that. As well as alot of other interisting questions. worth a read

[Joe]

hard to read the broken German-Engrish translation but I do think that Watercooling in general is moving back to the niche. Its play time in the mainstream is well on its way to becoming just a fun memory of the mid 2000's.

[ben333]

Well unless they make processors at a low enough wattage so it really doesn't matter whats cooling it I think water cooling will always be better than air... Has anybody taken one of those air sinks, cut the heat pipe ends and pumped water through them? That'd be an interesting idea...

[billbartuska]

The writing is on the wall........

As transistor gate technology improves, electron migration becomes more and more of an issue. Intel's current line of 45nm dies have "tuned" gate manufacture to the point that additional cooling is having less and less effect on overclockability.

Anandtech:
The last seven generations of Intel processors have utilized traditional metal oxide semiconductor (MOS) materials, namely silicon dioxide (SiO2) and other polysilicates. This has spanned a period starting with the Pentium, originally built on 0.6 micron (600nm) node technology, all the way to Core 2 Duo/Quad parts built on 65nm. However, for the first time in over 40 years, Intel has significantly changed some of the basic components used in transistor fabrication. Hafnium dioxide (HfO2) has replaced SiO2 as the gate dielectric material, along with other carefully chosen new metals, for use in the formation of gate electrodes in both PMOS and NMOS transistors. These new materials, along with the right process recipe, have cut gate leakage by a factor of more than 10x while simultaneously delivering an astonishing 30% decrease in transistor switching energy. The result is a cooler running, more energy efficient, and high-performance processing powerhouse. Moving to a smaller node process technology allows for the potential of two things to happen. The first is the ability to make smaller production dies, reducing manufacturing costs, and ultimately making products more affordable to the consumer. Alternatively, designers can take advantage of the increased transistor density made possible by the new process and develop next-generation solutions that pack even more transistors into the same space as before. Smaller transistors also allow Intel to take advantage of lower switching energies up to a limit. This is true until smaller and smaller transistor gate dielectric boundary layers create conditions in which *power leakage - even with the transistors in the "off" state - become excessive. From this point on something will have to change if Moore's "Law" is going to continue.

* Read "electron migration"

So, pretty soon we will have CPUs that run cool, but are not overclockable without rapidly being destroyed.

[Jag]

Originally Posted by Joe

Its play time in the mainstream is well on its way to becoming just a fun memory of the mid 2000's.

Well, the community couldn't cope with press releases every month or so.
It was simply too much.

The "heat" factor is still there, (9800GX2, anyone?), and aircooling isn't able to handle it adequately.

On the other hand, in my opinion Wc, still has to be more specific, ie, specialized, like low-noise, better integration, etc, or simply follow the motherboard manufacturer's game, of selling boards based on massive PR actions and how much it can overclock.
After the initial momentum, it slowly fades away.

BillA, a long time ago mentioned the way AMG developed parts for mercedes, (a niche market), sort of implying that the cooling manufacturers, maybe, should progess in that direction.
It's a thought worth remembering.