Noctua NH-D14 vs Corsair H100 for 2600K OC'ing? Update: Went with both!

[jvroig]

braggadocious

I was surprised to find this is actually a real word. Learn something new everyday!

Here's another data reduction analysis for you, interesting that there is a nice wide clockspeed plateau in which the performance/watt basically remains flat (so going for more performance by way of higher clockspeeds does not sacrifice performance/watt):

4GHz seems like an obvious choice vs stock, since no performance/watt hit takes place. 4.5-46 is "safe" territory. Personally, I would settle only for 5GHz. Why bother buying the highest-end board, power supply, and cooling only to settle for a wuss overclock that my peers with much cheaper boards and/or cooling will accomplish anyway?

To paraphrase something I heard/read from Donald Trump: "Go 5GHz, or go home!" :thumbsup:

 

[Idontcare]

4GHz seems like an obvious choice vs stock, since no performance/watt hit takes place. 4.5-46 is "safe" territory. Personally, I would settle only for 5GHz. Why bother buying the highest-end board, power supply, and cooling only to settle for a wuss overclock that my peers with much cheaper boards and/or cooling will accomplish anyway?

To paraphrase something I heard/read from Donald Trump: "Go 5GHz, or go home!" :thumbsup:

LOL, that's my feeling too

When I realized I was shelling out $360 for the mobo alone, and yet the CPU was a mere $315, if felt really odd and like the world was out of balance. Then the SSD was $500 and I was like "damn, Intel's in the wrong business". (no, not really, but you get my point, the pricing is kinda wonky relative to the performance contributions each component makes)

That said, it does feel a little wierd to be shoveling 1.5V down my 32nm CPU when I killed my 65nm CPU by shoveling 1.6V down it. Although I readily admit this might be one of those cases where statistics and sample size (i.e. "1") are the dominating factor in the data.

 

[aigomorla]

4GHz seems like an obvious choice vs stock, since no performance/watt hit takes place. 4.5-46 is "safe" territory. Personally, I would settle only for 5GHz. Why bother buying the highest-end board, power supply, and cooling only to settle for a wuss overclock that my peers with much cheaper boards and/or cooling will accomplish anyway?

To paraphrase something I heard/read from Donald Trump: "Go 5GHz, or go home!" :thumbsup:

uhhh... yeah gillbot made that term an adjective and calls it getting aigo'd.

And getting aigo'd means doing all you just listed with adding water on top..

 

[Piano Man]

one more thing u guys...

in contrast to tim.

i dont use the best... because the best has issues applying.

The best is indigo extreme... no contest..
HOWEVER...

it doesnt work well on liquid cooled systems, especially non presealed kits.

You need a load time on it in temps near 60C. This just isnt possible on a LC'd system unless your pump is turned off, or your fans are off.

After that to my knowledge and after many many many testing, the next best by far is Prolimatech PK-1. It has the most consistant mounts from testing on skinnee labs as well as my bench.

Then after that, go for Artic Cooling's MX-1. Again... its super easy to apply.



Once again, the laws of TIM isnt whats the best... its what u can apply the easiest and with the most consistency!

IX is a piece o cake to install on Air setups. The only difficult part is watching your CPU go to 95-97C during the reflow procedure.

 

[aigomorla]

IX is a piece o cake to install on Air setups. The only difficult part is watching your CPU go to 95-97C during the reflow procedure.

i panic if i see 60's on my cpu.

and no im not kidding u can ask idc..

i dont think he's ever seen me publically hit 70C's :hmm:

A Watercooling diety does not fall into hell! err...

I mean it would be bad to see the top tier water system on this forum not able to do anything like that..

 

[Idontcare]

Now that I wrapped up my lengthy analysis of the power-consumption attributes of my 2600K, its time to turn my attention back to evaluating the NH-D14 versus my H100.

Heh, downside of exposed copper - it corrodes Statue of liberty is green for a reason Had to buff-off the patina on my previously lapped NH-D14 before mounting it to the 2600K. (the anaerobic environment provided by the TIM prevents corrosion while mounted, its only an issue if you lap your HSF and then leave it sitting out for a few weeks exposed to the air)

Compared to the H100, installing the mounting bracket for the NH-D14 is a quite a bit more involved

For what its worth, the mounting system itself is quite robust as needed to manage the extra weight.

Putting it all together was no problem. The problem for me came from my MIVE-Z though once I started doing my temperature tests - the ASUS qfan stuff through the BIOS does not work with 3-pin fans So my Noctua fans were running full-bore even when the CPU was clocked at 1.6GHz :|

That's no good. So how loud are those two fan?

48.2 dB at a distance of 18" from the HSF. For comparison, the H100 with four noiseblocker fans all running full-out is 46.7 dB.

Just to check performance of the noctua fans at full rpm, I ramped up the computer to 5GHz and loaded IBT. Temps peaked at 93C, this will be the reference point for comparing the "plan B" solution that will be detailed below.

I needed PMW fans on the NH-D14 so that the noise versus temperature profile based on the fan controller on the MIVE-Z would be used. Otherwise there would be no good way to make a comparison to the H100. So off came the Noctua fans and on went two of the noiseblocker multiframe fans. (the same ones I used on the H100).

I've got data and pics but it'll have to wait till tomorrow, too late here to do more now.

 

[RussianSensation]

Just to check performance of the noctua fans at full rpm, I ramped up the computer to 5GHz and loaded IBT. Temps peaked at 93C, this will be the reference point for comparing the "plan B" solution that will be detailed below.

Based on this graph, the H100 was also hitting low 90s with noise ~ 46.7 dB

Noctua $75 vs. $160 H100 setup. Oh-oh, I feel like there may be an upset in the making once those noiseblockers are mounted on the NH-D14. Either way, it's already impressive that NH-D14 keeps up with a 3dB higher noise level penalty but costs less than half!

 

[Idontcare]

Here's one graph depicting the comparison:

While it is hard to see the differences, which itself is a notable observation, the basic story here is that the NH-D14 tends to run just a smidgen warmer than the H100 as well as being a smidgen louder.

The price/performance here is clearly in favor of the NH-D14, there's no question.

Of no surprise, since the operating temps are just a bit higher, the minimum voltage had to be increased by the tiniest amount (0.005V) for just a few of the clockspeed points on the curve.

I'm also at a loss to robustly explain why the noise levels emanating from the NH-D14 with two noiseblocker fans are actually higher than the noise emanating from twice as many (four total) of the same fans when those fan are bolted to the radiator of the H100.

All I can surmise is that the fin arrangement on the NH-D14 is either causing more noise because of vortices's and so on that form, or that the H100 radiator simply does a better job "buffeting" the noise such that it simply is quieter.

Subjectively, standing in the same place while the system was running the different coolers, the NH-D14 was perceptibly louder despite having half as many fans involved.

(recall that sound pressure doubles for every 3dB)

More coming.

 

[RussianSensation]

All I can surmise is that the fin arrangement on the NH-D14 is either causing more noise because of vortices's and so on that form, or that the H100 radiator simply does a better job "buffeting" the noise such that it simply is quieter.

Can you please create a separate graph depicting NH-D14 with stock fans vs. Corsair H100 with 2 stock fans, up to 5.0ghz? That would also provide us with an idea of how the 2 compare in the 'out of the box' scenario.

 

[Idontcare]

Can you please create a separate graph depicting NH-D14 with stock fans vs. Corsair H100 with 2 stock fans, up to 5.0ghz? That would also provide us with an idea of how the 2 compare in the 'out of the box' scenario.

I can do "something" like that, but I destroyed any "out of the box" comparisons to be had when I irreversibly altered the two by lapping them and my 2600K.

Comparing the noise from the two with there stock fans is doable but I can already tell you who wins - NH-D14 for noise and the H100 for cooling. We've got plenty of web reviews that show this which is why I didn't give it much thought to just move on to the next step as I did.

I definitely see the value in NH-D14. I don't see the value in the stock H100. It cost so much $50 more and yet best case scenario gives me maybe 2-3C better temps at the expense of much more noise.

The better temps do result in a minor 0.005V Vmin difference, but 0.005V doesn't gain you a bump-up in the multiplier for OC'ing purposes (it's about 1/4 to 1/5 what is needed to juice an extra 100MHz out of the CPU).

I don't regret buying the H100, the hobby experience I got out of running these tests was more than worth it from an entertainment/dollar perspective. But I won't be buying another one. Kinda the same conclusion I had from my VapoLS. More novelty than value in comparison to air-cooling.

But yeah, I run a few more tests to put into numbers what I've put in words here.

Want to upload my Indigo Xtreme results too (cliffs: on the D14, inigo dropped the temps ~1C from where NH-T1 was getting me...is it worth $10 for that 1°C? maybe, but probably not)

 

[Idontcare]

Oh lord I'm such an idiot...somehow I managed to delete all my screenshots for the Indigo Xtreme temperature ramp installation

I did manage to avoid deleting the a few photos though. Here's the tim shot after removing the NH-D14 when it had NT-H1 tim on it:

Here's the Indigo Xtreme setup on the CPU, prior to the thermal ramp:

 

[aigomorla]

Oh lord I'm such an idiot...somehow I managed to delete all my screenshots for the Indigo Xtreme temperature ramp installation

lol idc!

NOOOOOOOOOOOOOOOOOOOOOOOB~

*gonna get ban'd from the supermod now..* :X


Can u try to recover with software? or is it gone for good as in no dorthy u aint coming back home to kansas?

 

[Idontcare]

lol idc!

NOOOOOOOOOOOOOOOOOOOOOOOB~

*gonna get ban'd from the supermod now..* :X


Can u try to recover with software? or is it gone for good as in no dorthy u aint coming back home to kansas?

Deleted on SSD...well, from one of them anyways. I really have no recollection which one I had them stored on at the time, which is probably how I ended up deleting them because I assumed the one's I was deleting were superfluous copies.

I did find my IR gun photo that I took showing how hot the NH-D14 was getting during the fan-free thermal cycling to set the Indigo Xtreme.

The red light you see in the center of the face of heatsink finks is the laser pointer built into the IR gun to show where the temperature is being measured at. Pretty cool little toy.

At any rate, the Indigo Xtreme install went exactly as the instructions said it would go.

Now, I need to take out the Indigo and go back to NT-H1 because I forgot to do a few tests with the NT-H1 on the NH-D14 before I go back to the H100 to test the Indigo Xtreme with that one.

Oh, I wanted to put this up:

Looking at Case #3 in comparison to Case #2, it seems I was mistaken above when I quipped that the Indigo merely improved load temps by a degree or two.

The Indigo Xtreme actually improved temps by 3.7°C over those with the NT-H1 TIM when the 2600K as running at 5GHz. That is impressive.

I also ran the conditions that Russian was interested in, the performance of the NH-D14 with the stock fans (see Case #4) and I also added in the performance when using the stock 120mm fans that come with the H100 (Case #5).

I plan to run the cross-test with the H100 as well.

Notice that the stock Noctua are just a little bit louder than the Noiseblocker, the noise difference is perceptible, you can easily tell the difference in the two by ear.

The corsair fans, ugh, how to put this - they move a lot of air (notice the extra 3.5°C temp reduction from Case 3 to Case 5) but holy moly are they LOUD.

The difference between 46 dB and 57 dB is the difference between :awe: and D:

More tests later...but last chance to request any more tests with the NH-D14 while its got the Indigo Xtreme TIM. These TIM patches are one-time use and cost $10 per application, so I won't be doing NH-D14 under Indigo again...its now or never if you want any more tests with this combo.

Otherwise I'll resume more testing tomorrow with the NT-H1 TIM as well as the H100 + Indigo combo.

 

[RussianSensation]

Great results!! Thanks a lot for going the extra mile and providing such a detailed analysis IDC.

I am very impressed by the improvements that Indigo made, but it is pricey.

It looks like NH-D14 is keeping up with your Corsair H100 at similar noise levels @ 5.0ghz.

 

[Idontcare]

Yeah, its fun :thumbsup:

I am very impressed by the improvements that Indigo made, but it is pricey.

I look at this way - 1 tube of NT-H1 cost me $10, 1 application patch of Indigo Xtreme cost me $10.

(you get two patches in that $20 kit, plus they include a bottle of the xtreme cleaner)

Definitely if you are in the habit of reseating HSF's on your CPU then its not the way to go. But wow, the stuff really works, doesn't it?

It looks like NH-D14 is keeping up with your Corsair H100 at similar noise levels @ 5.0ghz.

Oh yeah, the NH-D14 is not holding me back at all. The TIM makes a bigger difference in cooling performance than the difference between the NH-D14 and the H100.

Pretty cool.

Any more tests you care to see before I pull the NH-D14 and lose my Indigo patch?

 

[Idontcare]

Since I have already invested a fair amount of energy into determining how to correctly re-normalize sound levels to account for varying ambient sound levels (you don't just add and subtract like you can with temperature), I figured I'd share the formula here for the benefit of anyone who finds themselves wanting to do this at some point,

Let's say you measured the sound levels of a HSF under to different sets of conditions - maybe two different fans - and you wanted to make sure you could compare those numbers in apple-to-apple fashion.

To do this you need to renormalize one of the sound level values to account for the changes in ambient noise (if any such change has occurred) between the collection of data for the two test cases.

For an example to speak to, let's say you measure the sound level of Fan X to be 43.3 dB and the ambient sound levels (fan at 0 rpm) was measured to be 39.5 dB.

And let's say you measure the sound level of Fan Y to be 43.2 dB and the ambient sound levels (fan at 0 rpm) was measured to be 38.7 dB (it's quieter for some reason when you tested Fan Y).

If you ignore the ambient sound levels your conclusion would be that Fan X is louder than Fan Y.

But we see the ambient for Fan Y is lower than that of Fan X, meaning Fan X's sound level is higher because there is more background noise not coming from the fan.

So we need to "renormalize the ambient" for one of the two tests such that they are both based on the same background noise level.

The equation to do that is:

Let's renormalize the test results for Fan Y to the ambient levels of Fan X (ambient was 39.5 dB):

Now we see that once we've correctly accounted for the difference in background ambient sound levels, the sound level for Fan Y is actually 43.5 dB and as such it is "noisier" than Fan X at 43.3 dB when both are renormalized to a background ambient sound level of 39.5 dB.

If you use Excel for spreadsheet analyses, this is the formula in Excel:

=10*LOG(10^(B1/10)+10^(C1/10)-10^(D1/10))/(LOG(2)+LOG(5))

Where B1, C1, and D1 correspond to the cells that contain the data described above as A, B, and C.

 

[mrjoltcola]

Hey IDC, I'm wondering, is the bench mount comparison fair to the H100, since you aren't doing it inside a case? Running it on the bench = unrealistically low ambient temps for the HSF. (I didn't read the full thread, so if I missed something related, sorry).

EDIT: Nevermind, I saw the pic of the board on the bench and jumped to the wrong conclusion that you were benching outside the case. Sorry. I actually went back and read the whole thread.

Great thread as usual, IDC!

 

[Idontcare]

Yeah the "bench" is outside the case for these tests so far.

The case it will go into is the Black Pearl (build thread here), which has an open-mesh siding concept. The air flows in and out so freely, I am curious to see how of an impact the case does have, that test will happen too.

^ because the case is so open, that is why I am actually putting so much energy into the noise aspects of the HSF. The fan noise doesn't get the usual amount of dampening from the case sides.

 

[Idontcare]

I took the NH-D14 off which had the Indigo Xtreme, there's the before mount and after mounts shots.

Before:

After:

The Indigo Xtreme leaves behind a thin metal film that can kinda be peeled off, but not very easily. Here is what was left on the CPU and the NH-D14 after I did my best to remove it with "conventional" means:

This was after using IPA and the Xtreme cleaner provided with the Indigo Xtreme kit.

To remove the rest of the metal I used a razor blade, but then I needed to repolish the surfaces to get rid of the scratched from the blade.

Shots after repolishing:

 

[Idontcare]

After I cleaned up the Indigo mess, I put NT-H1 back onto the NH-D14 to rerun some of the 5GHz tests.

Here are the results:

I took a photo of the NT-H1 dallop before mounting (next to a grain of rice for scale), as well as a photo after I unmounted the NH-D14 once I was done with the test above:

 

[Tikerz]

This has to be the most in-depth review/comparison of anything in any hobby that I've been apart of in my entire life.

 

[RussianSensation]

This has to be the most in-depth review/comparison of anything in any hobby that I've been apart of in my entire life.

:thumbsup: IDC is making the professional cooling reviews amateur in comparison. :thumbsup:

 

[mnewsham]

:thumbsup: IDC is making the professional cooling reviews amateur in comparison. :thumbsup:

Someone call Anand

 

[Idontcare]

LOL, thanks guys for the kind words :$

I thought I would try another one of those "myths about thermal paste" tests...this time it was with toothpaste! :twisted:

^ yes that is toothpaste on my 2600K

The results are at 4.5GHz:

Ouch! Yeah, toothpaste as a TIM is truly NOT a good idea!

 

[mnewsham]

can't say I'm not surprised on the toothpaste front... :biggrin: