• Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!

air vs. watercooling vs. anything else

Overclockers is supported by our readers. When you click a link to make a purchase, we may earn a commission. Learn More.

cujo

Member
Joined
May 29, 2003
Location
Ontario, Canada
if i crank my sf II to max rpm, my load temps drop to 30C at 1.92vcore. haven't tried to oc at that voltage but i was doin a burn-in at the time just to see if it would help.

anyway, my real question is: as long as your temps remain low, is there any difference between different types of cooling? like, if i put my computer in a freezer, as is, with my air cooling and temps were around 0C or lower, would some other high end cooling like a mach II get me any different overclocking results?

or would a watercooling setup with temps at 45C be any different than an air cooling setup with temps at 45C? i would think no.

this is probably a stupid question but so many ppl have water cooling. the only real benefit i can see is so keep temps low in a room temp environment with low noise.
 
Water cooling is usually used for extreme overclocking where people have the best components already setup, then watercooling usually squeezes the last few MHz out of it.

The other reason would be noise.
 
from my understanding is if you can hit below 50 with air and you get the same temps with water.....then stick with air for portability reasons....the phase changing is a different story I think.......seeing as electrons move faster at lower temps...there is less resistence....ala a much nicer overclock.....the only large diff I see from going from high end air to water is sound.....the temps to me are negligable......I went air cooling for practicality since I like to lan......and price.....I set the fan speed low.....mainly cause I can't stand the loud noise.....and also there is such a little gain from 2.3Ghz to 2.5.....I hope I answered your question..my mind is everywhere right now:D
 
heh, should treepop post while he's high? Yes, I say.
 
thanks. so it is just temps like i thought. ok so if i hit a roof but still have load temps in mid to low 40s then i'd probably get no higher even with extreme cooling.
 
As far as CPU speed, it does not matter whether there is a fan or water running above it, as long as there is enough voltage and cooling to keep temperature below certain point so it can run stable at certain frequency.

Some related posts about various type of cooling, cost and effect on CPU overclocking.

hitechjb1 said:
Overclocking limit and cooling cost for different types of overclocking:

When running at full load (e.g. Prime95) to test for stability by adding small step of Vcore (e.g. 25 mV), if the increase in CPU frequency per CPU temperature increase is more than 10 MHz / C, I think there is still have some more overclocking to go, for a given cooling setup. For extreme cooling, one could use 12 MHz / C instead of 10 MHz / C.

If it is less than 10 MHz / C, it is beyond the break-even point or point of diminishing return, it would be getting harder and harder to overclock.

At 2-3 MHz / C, it reaches the limit. Then a better cooling would be needed, such as a better HSF with a smaller thermal resistance C/W, or going to water, or extreme cooling.

The above rules can be applied to various type of coolings, including air, water, extreme cooling

But usually, going to a better cooling becomes less and less price-performance efficient, as shown below, except for absolute higher MHz and environmental reason, noise reduction in the case of water cooling.


The cost in terms of overclocking MHz for various cooling, using Tbred B 1700+ DLT3C as an example:

- stock air HSF: from 1500 to 2100-2200 MHz, 60 C
$10 for 47% overclocking => ~ 4.5% / $ or 60-70 MHz / $ (best price)
Recently, I tested a new AMD stock HSF, I got 2300 MHz at 65+ C, fan at 3100 rpm (very quiet)
For actual MHz per $, it is 210-220 MHz / $.

- high performance air (e.g. SK-7 / SLK-800U): from 1500 to 2400-2500-2600 MHz, 50 C
$30-40 for 60% overclocking => ~ 2% / $ or 30 MHz / $ (best price-performance tradeoff)
For actual MHz per $, it is 60 - 80 MHz / $.

- DIY high end water: from 1500 to 2500-2600-2700 MHz, 30-40 C
$150-200 for 80% overclocking => ~ 0.5% / $ or 7 MHz / $ (not price-efficient, but good for the ears)
For actual MHz per $, it is 14 - 18 MHz / $.

- extreme cooling: from 1500 to 2700-3000 MHz, -10 to 20 C
$300-500-1000+ for 100% overclocking => 0.1 - 0.3% / $ or 2 - 4 MHz / $
For actual MHz per $, it is 3 - 10 MHz / $.

Without going into each individual type of extreme cooling, and the still evolving setup such as thermoelectric, chilled water, phase change, liquid nitrogen, ... , the cost per MHz for each type of extreme cooling is very high.

Observation, the MHz/$ is roughly about 2-4 times more costly as moving up the cooling chain each step from stock air HSF, to high performance air, to water, to various exterme cooling. The improvement from stock HSF to high performance HSF (such as copper) is noticeable (~ 15-20%) and worth the added price. The gain in MHz from high performance air to water is minimal (100-200 MHz), except for environmental reasons such as noise reduction, also water cooling requires more setup time and more maintainence time.

The numbers used are for illustration purposes to show the trend and the cost involved of the various types of CPU cooling. The exact MHz and $ amount may vary by a slight amount and would not change the overall picture and general trend.


Some numbers to determine max CPU overclocking frequency - Vcore vs temperature,
When do the CPU's slow down?
(page 13)
Explanation (page 13)

Overclocking limit and cooling cost for different types of overclocking(page 13)
 
Back