Sign in with
Sign up | Sign in
Your question
Solved

Do i need water cooling for i5 3570k OC

Last response: in Overclocking
Share
June 14, 2012 1:35:47 PM

I will be doing some OCing on my 3570k down the road, and right now i dont think it will be very extensive OCing but instead of buying and aftermarket fan should i get water cooling?

More about : water cooling 3570k

Best solution

June 14, 2012 2:23:36 PM
Share

Water cooling if fun and yet kind of expensive and can be very frustrating at times. It is a big undertaking. With water cooling though you will see lower max temps but your idle temps won't change much. What case do you have? What do you mainly use your PC for? Would you want to water cool your GPU(s) also?

For me I set up water cooling because my 2 6970s in CrossFire soundled like a jet engine when under load so I went water cooling for a quiet pc.
a c 324 K Overclocking
June 14, 2012 2:24:43 PM

huntx23 said:
I will be doing some OCing on my 3570k down the road, and right now i dont think it will be very extensive OCing but instead of buying and aftermarket fan should i get water cooling?


Based on this, no, you don't. Just get a decent air cooler (or even use the boxed Intel cooler) and you will be fine.
Related resources
June 14, 2012 2:46:47 PM

alright i think ill stick fans (less hassle and less complicated) and since this is my first build i dont want to get too crazy.

i think buying a case will good cooling will also help
June 14, 2012 3:03:33 PM

Best answer selected by huntx23.
a c 324 K Overclocking
June 14, 2012 3:17:29 PM

Those Corsair coolers aren't going to give you any better cooling performance than a good air cooler.
June 14, 2012 4:23:21 PM

rubix_1011 said:
Those Corsair coolers aren't going to give you any better cooling performance than a good air cooler.


That's not true... They offer lower max temps than even the best air cooler and it is quieter, especially if you install after market fans on it. Also the more you overclock it, the bigger the difference.

http://www.pcgameware.co.uk/corsair-h100-review/
a c 324 K Overclocking
June 14, 2012 4:41:26 PM

I disagree. You aren't going to see anything more than 1-2C lower temps (at the expense of louder fans...that you have to pay extra for) in order to accomplish these minimally lower temps. The only one that really is able to do this in a best case scenario is the H100 and only because it's a 2x120 rad. It's still a very low flow pump, high FPI rad and needs push/pull fans set to 2000rpm or higher to accomplish this.

All of these items combined are a fail. If you are paying in the neighborhood of $115 for an H100 and then have to spend $10 a fan (x4 for push/pull for good fans), you should have opted for an entry level Rasa or Raystorm watercooling kit and benefited from real watercooling.
June 19, 2012 7:18:30 PM

rubix_1011 said:
I disagree. You aren't going to see anything more than 1-2C lower temps (at the expense of louder fans...that you have to pay extra for) in order to accomplish these minimally lower temps. The only one that really is able to do this in a best case scenario is the H100 and only because it's a 2x120 rad. It's still a very low flow pump, high FPI rad and needs push/pull fans set to 2000rpm or higher to accomplish this.

All of these items combined are a fail. If you are paying in the neighborhood of $115 for an H100 and then have to spend $10 a fan (x4 for push/pull for good fans), you should have opted for an entry level Rasa or Raystorm watercooling kit and benefited from real watercooling.


The problem with ivy bridge CPUs, and how their built is that they have high-spikes of heat that aircoolers can't adjust fast enough to compensate. By the time air coolers ramp up to speed it's already too late and the heat spikes too high. (You'd know this if you'd read Tom's review on the Ivy Bridge architecture)

The closed loop coolers and real watercoolers are able to deal with the spiking problem with ivybridge CPUs due to the nature of water and how it transfers heat.

So where your assessment is generally correct with CPUs like the 2500k, and the 920 i7 from the X58 systems, it is incorrect for ivy bridge.
a c 324 K Overclocking
June 19, 2012 7:22:20 PM

The 'heat spikes' are the same for any CPU on any cooler. I see them on my full watercooling loop when running load-intensive applications. It's a matter of how much the cooler can handle the head loads.

Edit/Add:
Quote:
The problem with ivy bridge CPUs, and how their built is that they have high-spikes of heat that aircoolers can't adjust fast enough to compensate. By the time air coolers ramp up to speed it's already too late and the heat spikes too high. (You'd know this if you'd read Tom's review on the Ivy Bridge architecture)

I read this, but don't recall where this was stated...can you link to the page? I'd like to go back and re-read this as I disagree if this was actually said.

Furthermore- the temps listed by CPU monitoring isn't the temp that the actual cooler is handling at that very second; it's the temp reported by the sensor inside the CPU IHS, and fluctuates many times a second. Therefore, the temps being reported as 'spikes' aren't what the cooler is actually attempting to dissipate at the instant they are reported, yet the CPU thermal threshold is dependent upon what these reported (or actual) temps are.
June 19, 2012 8:12:30 PM

SB processors used soldered heat spreader, where IB uses a thermal paste.

This is causing thermal-throttling to occure before the aircoolers have a chance to ramp up speed enough to compensate.

People who replace the built in thermal paste with the solder are able to achieve 5.0ghz on air.

The article recommends closed loop or real watercooling to prevent thermal throttling.

I have tried 3 different solutions on my IB system, Air, H100 and now a 360 rad watercooling loop. There was a huge improvement in performance from Noctua D14 to a H100. Temps dropped a bit using a real watercooler, but it's obvious that the triple 360 rad was overkill for just my CPU (But I intend to GPU watercool as well)

http://www.tomshardware.com/reviews/ivy-bridge-overcloc...

Read the article, it explains it.

You can disagree all you want, but you're still wrong. In older CPUs, you would have been correct, but IB is physically different and has different quirks to work around. You just didn't have all the facts. Now you do.

Also, it's not something easy to monitor, even the tech in the article said he had to use a script to catch it.
a c 324 K Overclocking
June 19, 2012 8:21:28 PM

Very interesting, and reading that does explain it more...just quite different than what I would expect.

However, to look at this further, any air cooler would also be at a thermal threshold and dependent upon fan RPM in order to actually be any more effective where someone would assume that within a split second, the CPU heatsink itself (being the mass of copper and possibly heatpipes) would be able to absorb the heat load changes during the time between the load occurred and the fans spun up to cool the heatsink. Apparently, there is such a large surge of heat at those clock speeds that it completely saturates the CPU cooler and causes thermal throttling before the CPU cooler can actually attempt to dissipate the heat being produced.

Weird. Even the fairly mediocre Corsair H-series would be at advantage here with the ability for this thermal load to be moved into the coolant stream more effectively than ambient air would for an air cooler.

I stand corrected on my previous statements.
June 19, 2012 8:34:13 PM

Ya I normally never bothered with liquid cooling other then for looks. My new IB build on air was causing me issues, and after reading the review here I followed the suggestion of the H100 and it helped a huge amount.

But I have a feeling this isn't going to be a normal thing. The next chips will probably go back to the solder, and this is probably just a fluke, and $50 - $75 air coolers will work just as good as H100.
a c 324 K Overclocking
June 19, 2012 8:38:51 PM

Honestly, it's a very interesting conclusion and one that I didn't see coming, especially since I just went to SB. Perhaps this was a cheaper and simpler solution for IB? Either way, it was just strange to see a significant jump from SB to IB, taking into account the +2 cores.
!