Overclock.net banner

1 - 20 of 20 Posts

·
Registered
Joined
·
86 Posts
Discussion Starter #1
All data was collected by averaging the RealTemp max value of all six cores over a 15 minute run of Prime95 v26.6 (it generates a more consistent CPU temp compared to other stress tests IMO), and then subtracting the ambient temperature as measured near the chassis intake. The only exception to this was the 140 Power test which was terminated after 5min due to high temperatures, but past experience with that cooling setup has shown that equilibrium is reached in a couple minutes compared to water cooling setups that have much more thermal mass and require 10 minutes or more to warm up. Both RealTemp and my ambient thermometer are incapable of measuring fractions of a degree, so I would estimate a ~2° margin of error.

5820k 4.4ghz at 1.3 Vcore 1.9 Vinput Prime95 v26.6 15min

44.5° C delta Custom Loop 280mm 1600 rpm
47.8° C delta Custom Loop 280mm 1100 rpm
48.2° C delta H110i GT (Vardar fans) 1600 rpm
48.5° C delta H110i GT (stock fans) 1600 rpm
57.5° C delta TRUE 140 Power ty-143 fan 2500 rpm

I went into this test with a preconceived notion that the Corsair unit was going to suck, and it ended up impressing me instead. Its cooling performance comes close enough to the custom loop as to make no difference in my overclock, though someone willing to push 1.4+ Vcore 24/7 might need the improvement. I didn't bother testing the Corsair unit at full fan speed because it was obnoxiously loud, and I wanted to test at a fan speed similar to the F2-140 fans I selected for the custom loop. With the Vardar fans attached to the H110i GT, I found no temperature difference at the same rpm, but the Vardar fans were significantly quieter. The Corsair unit's only major downside is how annoying the stock fans are, but after adding ~$40 shipped for a pair of quality fans, you end up with a total cost approaching the EK Predator or Swiftech Prestige series. The pump on the Corsair unit was also a bit noisy for my taste; even on "quiet" setting, it was still significantly louder than my soft mounted D5 Vario on setting 4.

I still think the 140 Power is a great cooler, especially for quad core CPUs (it had no trouble cooling my 3570k at 4.7ghz 1.375 Vcore with the included 1300 rpm fan), but my 5820k at 1.3+ volts seems to overwhelm it. That said, a small 100mhz drop in frequency would make the 140 Power viable on my 5820k, and on a performance/price basis, it still comes out on top by miles. Anyone building a system on a budget with an eye for gaming performance should take a good look at the ~$50 heat-sinks since the money saved here is almost always better spent on a strong GPU.

Here are the relevant parts used for my custom loop:

HWLabs Black Ice 280 GTS X-Flow radiator
EK Supremacy EVO X99 block
EK D5 Vario pump
EK Vardar F2-140 fans

With the GTS radiator, this is obviously a loop geared for low-rpm operation. I probably could have squeezed another degree or two of cooling performance at 1600 rpm with a higher fin density radiator, but it would have meant sacrificing a significantly larger temp difference at the nearly inaudible 700 rpm speed I run my fans under normal loads.

I still have my 3570k and z77 board (though the x16 PCI-E slots are dead, hence the upgrade), so I might test the 140 Power against the H110i GT on it if there is any interest.
 

·
Gamer
Joined
·
1,244 Posts
How was the case airflow setup compared to the aio/custom loop vs the air cooler? Just out of curiosity
 

·
Registered
Joined
·
2,024 Posts
yes, an almost 10`C difference between air & water makes me suspect there is an airflow issue.... resulting the TRUE not being able to fully "breath"

But, on the other hands, thanks for doing this... more people should be doing this
thumb.gif
 

·
Registered
Joined
·
86 Posts
Discussion Starter #4
Airflow to the heat sink wasn't an issue. Previous testing had shown that heat sink intake temperature was within 1 to 2° C of case intake ambient (see pic of temp probe used below) depending on case fan RPM. The test run where I got the 9° difference between the 140 Power and H110i GT was done with the case fans cranked all the way up (bottom fans are 2150 rpm GTs blowing straight into the heat sink)



 

·
Premium Member
Joined
·
6,636 Posts
Quote:
Originally Posted by RnRollie View Post

yes, an almost 10`C difference between air & water makes me suspect there is an airflow issue.... resulting the TRUE not being able to fully "breath"
Recently I've seen alot of debate about CLC vs Air Cooling but you might have hit the nail on the head with a major advantage of CLC cooling - That you always have good air flow since the fans/radiator are mounted directly on the case.
 

·
Registered
Joined
·
86 Posts
Discussion Starter #6
Quote:
Originally Posted by BinaryDemon View Post

Recently I've seen alot of debate about CLC vs Air Cooling but you might have hit the nail on the head with a major advantage of CLC cooling - That you always have good air flow since the fans/radiator are mounted directly on the case.
That can certainly be the case when someone fails to optimize their chassis airflow setup, but radiators can have their own disadvantages as well. In a chassis with a very restrictive grill blocking air flow into the radiator, an air cooler might actually have an advantage. As an example, I just propped my computer case up on a couple wood blocks to increase floor clearance and gain better air flow into my bottom mount radiator. This small change resulted in a ~2° C improvement at 1600 rpm fan speed.
 

·
Gamer
Joined
·
1,244 Posts
Quote:
Originally Posted by CastleBravo View Post

That can certainly be the case when someone fails to optimize their chassis airflow setup, but radiators can have their own disadvantages as well. In a chassis with a very restrictive grill blocking air flow into the radiator, an air cooler might actually have an advantage. As an example, I just propped my computer case up on a couple wood blocks to increase floor clearance and gain better air flow into my bottom mount radiator. This small change resulted in a ~2° C improvement at 1600 rpm fan speed.
here is what i see going on in this airflow setup, I don't feel that its totally optimized for an air cooler. I may be wrong but the air cooling guru @doyll would sure know
biggrin.gif


 

·
Registered
Joined
·
86 Posts
Discussion Starter #8
Quote:
Originally Posted by Cysquatch View Post

here is what i see going on in this airflow setup, I don't feel that its totally optimized for an air cooler. I may be wrong but the air cooling guru @doyll would sure know
biggrin.gif


The pic I posted just below that one shows the temperature probe I used to confirm that wasn't happening. I had also previously tested that configuration against one with the heat sink rotated 90° and noticed no difference. Doyll and I had a debate in his thread about this already.

Keep in mind, this is a 6-core being run at a significant voltage increase, so a typical 155x quad would show a much smaller difference in temp between the different cooling solutions. I actually think my results prove that the budget single tower cooler is still the best option for most people, even someone with a 5820k like me, if they could put the money to use somewhere else.
 

·
Gamer
Joined
·
1,244 Posts
what kind of case was the test run in?
 

·
Registered
Joined
·
86 Posts
Discussion Starter #13
Quote:
Originally Posted by Smanci View Post

I'd think having no exhaust fans is a big issue. Where and how did you mount the H110i?
The radiators were bottom mounted as intakes. Exhaust fans are not needed; they are a lot like adding a second set of fans to a cooler for push+pull. As long as you have enough vent area for your intake vans to exhaust out of, you are good. Also, I blocked off every opening on the front and bottom other than the intakes to prevent re-circulation.

Since I tested heat sink intake temp vs outside ambient and found a 1° C difference after 10min at full load, I know the heat sink setup was not held back by case airflow.

If a lack of exhaust was an issue, the water cooling setups would have been affected to a greater degree due to them having to fight against the front intakes which were not restricted by a radiator.
 

·
Registered
Joined
·
2,024 Posts
I think a 4c difference is quite decent between the h110 and the custom, noise level comparison would be good too.

I would also be interested to see your results with one of the dual tower air coolers on a 5820k, rather then the TS 140 which has half the aluminium of those and so half the amount of surface area. I think that matters on higher tdp chips whereas the TS compares well on quads because it has the same amount of heatpipes and the heatsink surface area isn't ever a problem.

I'd think on the 3570k the difference would be a lot smaller between them all, very interested to see that if you decide to do that for sure.

Thanks for posting anyway.
smile.gif
 

·
Registered
Joined
·
2,024 Posts
i concede that it is possible that the TRUE just isnt good enough to keep a 6-core OCd monster in check
Actually it still does a good job, but this is where CLC/AIO/Custom or a "better" Aircooler clearly has an advantage.

So, now that almost all data is known, thats sorted out
thumb.gif
 

·
Registered
Joined
·
2,024 Posts
I was surprised that its such a huge difference as well tbh. Ive always been shy of recommending to people using higher TDP chips though for this exact reason, although I havent seen any real evidence until just now.

It would also be interesting to see results at say 1.1 volts going up in 0.05 increments on a hex or octa core to see at which point it really starts to lose it dramatically in comparison.

Similarly it would be interesting to see a comparison of the True vs say the NHD15 with the same fan(s) on such a chip. I wonder if the True would be almost on a level or even better up until the point where the heat produced starts to saturate the smaller heatsink as its obviously pretty good at getting the heat off of the CPU with its larger heatpipes then most.

Cryorig states that the R1 Ultimate has a TDP of 250 watts for example, Ive never been sure if that refers to the heatsink as a whole with the fans. As in if you were to whack on some TY143's instead does that increase the TDP?
 

·
Premium Member
Joined
·
9,788 Posts
Interesting test,however 15 mins is nowhere near enough for the water temps to reach their 'working temp'.
Do it again but for an hour,the temp gap will close slightly between the clc and air-cooler.

The smaller clc unit get heatsoaked quite quickly,not something you will see in 15mins...
 

·
Registered
Joined
·
86 Posts
Discussion Starter #18
The Corsair unit has a cold plate temp sensor that measures in 0.1° increments. I saw that temp hit its absolute max after about 10min, and it came within one degree of max in under 5min. It may take slightly longer for a system with a larger volume of water (my small loop only took about half a liter) and/or a lower CPU power consumption to reach equilibrium however.
 

·
Premium Member
Joined
·
9,788 Posts
Quote:
Originally Posted by CastleBravo View Post

The Corsair unit has a cold plate temp sensor that measures in 0.1° increments. I saw that temp hit its absolute max after about 10min, and it came within one degree of max in under 5min. It may take slightly longer for a system with a larger volume of water (my small loop only took about half a liter) and/or a lower CPU power consumption to reach equilibrium however.
The only temp that matters is the processor temp overall as it is the end game result,15 mins is not indicative of real use conditions for me.

I found the smaller h5/8/whatever to get overwhelmed quite easily,the h100 is something I have no experience in,nor interest if im honest.

I like the fact you put effort in doing a comparison tho,+ rep,they are time consuming and irritating to do so props to you fella.
 

·
Registered
Joined
·
86 Posts
Discussion Starter #20
Quote:
Originally Posted by B NEGATIVE View Post

The only temp that matters is the processor temp overall as it is the end game result,15 mins is not indicative of real use conditions for me.

I found the smaller h5/8/whatever to get overwhelmed quite easily,the h100 is something I have no experience in,nor interest if im honest.

I like the fact you put effort in doing a comparison tho,+ rep,they are time consuming and irritating to do so props to you fella.
I agree that the CPU temp is what matters, but a static cold plate temp is a good indication that the system has reached equilibrium. I referenced that temperature because it is measured at a 0.1° resolution, and also because it fluctuates less than the individual core temperatures.

The 120mm CLCs are in my view far worse than single tower heat sinks, and are only useful for builds that can't fit a decent air cooler. I had previously used an Antec 620 myself, and the TS140P I upgraded to was a significant improvement. I would have liked to test the Antec against the other coolers, but I unfortunately misplaced the mounting hardware. The 280mm CLC I tested by comparison has about 2.7 times the surface area of a 120mm unit, so it is easy to see where the difference in performance comes from.
 
1 - 20 of 20 Posts
Top