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

[Idontcare]

I feel the end is coming, there is light at the end of this long long tunnel!

OK, here's my final round of tests with NT-H1:

A few interesting observation here, Cases 1 and 2 are essentially comparing the H100 to the NH-D14 in as close to a "stock vs. stock" comparison as possible. (within the caveat that I lapped both mounting block surfaces, so they are not out of the box stock comparisons...but the web is replete with those kinds of comparisons already)

Stock comparisons at 5GHz, OC's are capable with the same Vcc, neither cooler enables a lower stable Vcc and the temps are quite comparable whereas the noise levels are not (in the favor of the NH-D14).

Equipping both the H100 and the D14 with the same fans, compare Cases 4 and 5, normalizes the noise levels and the temperatures, which makes the value proposition favor the NH-D14 (once again).

I also did a test to determine just how "noisy" the H100's waterpump is. I disabled all system fans, the only noise that could be heard was the sound of the water-pump. It has a distinct higher-pitch sound than that of the pans, but lower in sound level such that it is easily drowned out by fan noise when the fans are turned on.

That said, case 7 is the noise level, remember the ambient noise (computer turned all the way off) is 39.5dB, the water pump is running at 39.9 dB. Audible at 18 inches, but barely so.

To give you a feeling for how "quiet" these mid-40 dB sound levels are...my built-in dishwasher is rather quiet, and yet it sits 2 rooms away from my study, separated by two walls and two entrance ways. The level of noise that makes it from my dishwasher into my study makes the ambient noise levels rise from 39.5dB to approximately 42.5 dB.

The fan noise from my two laptops, individually, sends the noise meter to around 50 dB at 18 inches.

All these sub-50dB readings are quite quiet. I can easily discern between 45dB and 46dB, its not like I can't hear the rig, but the noise levels are amazingly quiet and easily forgettable.

My last set of tests will be to pair the Indigo Xtreme TIM with the H100. Results in the next couple of days

 

[deimos3428]

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:

If you're not too busy or tired of this project already, I'd be interested in seeing results without TIM.

 

[RussianSensation]

The toothpaste on sanded 2600k just made this thread truly EPIC!!

So,

Case 2 - air w/ stock fans: 63*C above ambient, 46.9 dB
Case 3 - water with 4 high-end quiet fans: 62*C above ambient, 47.2 dB

I honestly expected the H100 to do better than that.

 

[Idontcare]

If you're not too busy or tired of this project already, I'd be interested in seeing results without TIM.

You might be surprised by this, but I already did them Just didn't post it up.

Years ago when I lapped my Tuniq120 and Q6600 I tested without TIM and found it was every bit as good as going with AS5. I ran with AS5 anyways because it just felt like I was committing a sin to run w/o thermal paste.

This time around I naturally tested the same config (no TIM) with both the NH-D14 and the H100...only this time the results were disastrous.

At stock clocks, 3.8Ghz for my MIVE-Z, both heatsinks went to 98°C with IBT, throttled, and started downclocking the CPU to keep the thermals in check.

With both heatsinks, while under load and at 98°C I then backed off the mounting screws to relieve the pressure that might be causing the heatsink or CPU to bow and flex thereby creating an air-gap between the IHS and HSF.

Again for both heatsinks I went all the way to the point of just having the heatsinks loosely sitting on the CPU, the temps never got better.

I did not include those results in the tables because they are superfluous. 98°C is hardly interesting. But it can be said with data that toothpaste is definitely better than nothing (because no TIM sucks!)

The toothpaste on sanded 2600k just made this thread truly EPIC!!

So,

Case 2 - air w/ stock fans: 63*C above ambient, 46.9 dB
Case 3 - water with 4 high-end quiet fans: 62*C above ambient, 47.2 dB

I honestly expected the H100 to do better than that.

Me too, TBH

I am thinking though, after seeing what Indigo Xtreme did for the NH-D14, that Indigo with the H100 with 4 fans in push/pull might just net me out one notch lower on the Vcc curve for stable 4.9 or 5.0GHz operation, which would be a win for me.

But yeah...bolting a $220 cooling solution to my $300 CPU is not exactly exciting when the results are seen. There is something to be said of "balance"...the H100 + 2600K is not balanced, the H100 is excessive compared to cheaper and equally performing solutions.

Just goes to prove that turnkey watercooling toys are still just that. Albeit spendy ones.

 

[Jovec]

I am thinking though, after seeing what Indigo Xtreme did for the NH-D14, that Indigo with the H100 with 4 fans in push/pull might just net me out one notch lower on the Vcc curve for stable 4.9 or 5.0GHz operation, which would be a win for me.

I was going to post something like this yesterday, but decided to pass (I was also going to suggest a no-TIM test too). Unless the Indigo TIM allows you to do something you otherwise couldn't, then outside of the "coolness/best" factor it doesn't really matter. At least, that's the logic I used to talk myself out of buying some.

 

[RussianSensation]

Just goes to prove that turnkey watercooling toys are still just that. Albeit spendy ones.

I think there may be another explanation why the H100 didn't crush the NH-D14. If you take a look at the total power consumption, the 2600k @ 4.85ghz system is pulling about 140-150W less than a Core i7 950 @ 4.3ghz / X6 1100T @ 4.2ghz. It may be the case that an overclocked 2600k is simply not stressing the H100 enough to show that it's superior / i.e., NH-D14 is able to dissipate this level of heat without a problem.

If the comparison was done with a hotter CPU, I am confident the performance delta would grow in favour of the H100 as the air cooler's ability to dissipate a lot more heat may come into question. I think the H100 would have 'loved' to cool an overclocked GTX580

 

[jvroig]

Hey IDC, if you aren't completely tired from playing with your new toys yet, can you make one final mythbusters episode and tell us if "re-using the TIM is bad" is busted, plausible, or confirmed? Just unmount your HSF, pretend to do something for 5 minutes or so, then just remound the heatsink without cleaning out the old TIM and applying a new one.

If you did toothpaste and no TIM at all, you owe it to us, your demanding and discerning readers, to do a "reused TIM" test. Nay, you owe it to science!

 

[Idontcare]

I think there may be another explanation why the H100 didn't crush the NH-D14. If you take a look at the total power consumption, the 2600k @ 4.85ghz system is pulling about 140-150W less than a Core i7 950 @ 4.3ghz / X6 1100T @ 4.2ghz. It may be the case that an overclocked 2600k is simply not stressing the H100 enough to show that it's superior / i.e., NH-D14 is able to dissipate this level of heat without a problem.

If the comparison was done with a hotter CPU, I am confident the performance delta would grow in favour of the H100 as the air cooler's ability to dissipate a lot more heat may come into question. I think the H100 would have 'loved' to cool an overclocked GTX580

These tests are done at 5GHz, CPU is consuming >220W

If the H100 cannot shine under these conditions, I am not sure they ever will, nor can I argue a case based on heat transfer theory why it should.

Also consider the fact that water boils at 100°C and at 5GHz my CPU is reaching 94°C territory. Pretty sure the H100 is not designed to withstand the water temps reaching those levels, the tubes are made of plastic after all.

I personally feel much safer sitting next to an all-metal encapsulated D14 knowing the heat transfer mechanism itself relies on the boiling/evaporation cycle of a liquid. Worst case scenario for the D14 is that it gets hot and functions as designed. I can't say the same for the H100.

Hey IDC, if you aren't completely tired from playing with your new toys yet, can you make one final mythbusters episode and tell us if "re-using the TIM is bad" is busted, plausible, or confirmed? Just unmount your HSF, pretend to do something for 5 minutes or so, then just remound the heatsink without cleaning out the old TIM and applying a new one.

If you did toothpaste and no TIM at all, you owe it to us, your demanding and discerning readers, to do a "reused TIM" test. Nay, you owe it to science!

I like that idea :thumbsup: I'll run the test.

 

[Idontcare]

Hey IDC, if you aren't completely tired from playing with your new toys yet, can you make one final mythbusters episode and tell us if "re-using the TIM is bad" is busted, plausible, or confirmed? Just unmount your HSF, pretend to do something for 5 minutes or so, then just remound the heatsink without cleaning out the old TIM and applying a new one.

If you did toothpaste and no TIM at all, you owe it to us, your demanding and discerning readers, to do a "reused TIM" test. Nay, you owe it to science!

OK, here are the "reused TIM" results:

These were done 5GHz, this should represent the worst-case effects of failing to remove the old TIM and just reseating the mount.

With Cases 1, 2, and 4...the H100 orientation was as depicted in the following pic (the word Corsair on the waterblock is perpendicular to the dimms):

After generating the data seen in the table as "Case 1", I pulled the H100 and took the following pics:

I then reseated the H100 with the old TIM still in place on both the H100 and the CPU and reran the IBT test at 5GHz, peak temps increased by less than 1°C...which was surprising because this should be worst-case in that if I were running at lower clocks then the temperature delta would be even less.

I pulled the H100 and observed the following:

(I have to bust this into two posts for sake of picture count rules)

 

[Idontcare]

(the contents of this post relate to the data contained in the post above)

I then rotated the H100 waterblock 90° such that the word "Corsair" was now parallel to the DIMMs, and reseated the H100 with the old TIM still in place on both the H100 and the CPU.

Then I generated the data tallied in the table under "Case 3". Temperatures rose a further 2°C, inline with expectations of what I observed before in my testing with the H100 in which I found the "perpendicular" orientation to be better for getting lower temperatures.

I then pulled the H100 and took the following pics:

And for one final test, I decided to see how bad it would get if not only was someone to pull their HSF and then reseat it without replacing the old TIM but that they were also a clumsy goof who happened to not only stick his thumb into the TIM on his CPU's IHS but he also dropped some hair off his head into the TIM and left that behind as well, then proceeded to remount the H100:

One would expect this trifecta of disasters to reign terror on the cooling capabilities of the HSF...and it kinda did, peak temps rose to 97.7 (close to throttling but I confirmed it did not reach 98C and there was no throttling for the duration of the test) nearly 5°C higher than before...but there is hair in the interface for gosh sakes. D: I was completely surprised this thing did not bomb out on this test.

Thumbprints, hair, and reseats are not good for cooling, but they are also not the impending disasters that we tend to fear them to be.

 

[imaheadcase]

Has anyone tried flat on grease on cpu? Maybe some Goop. Try that, and see if the fire department has to come..

 

[Idontcare]

Has anyone tried flat on grease on cpu? Maybe some Goop. Try that, and see if the fire department has to come..

C'mon man, my mobo cost more than this CPU, I don't want to ruin my mobo by putting petroleum-based products like grease or Goop in contact with it. D:

 

[mrjoltcola]

ROFL

Now, IDC, can you please clock that up gradually to 6.3Ghz and let us know when the hair goes poof, and whether it is enough to trigger a motherboard fire??

 

[Idontcare]

I took some photos of the NH-D14 clearance around and over the heatsink fins on my G.Skill ripjaws x-series.

There is actually quite a bit of clearance above and beyond the height of the heatsink "fins" that extend above the dimms themselves.

 

[jvroig]

Thumbprints, hair, and reseats are not good for cooling, but they are also not the impending disasters that we tend to fear them to be.

So just plain old reseating without cleaning of old TIM and re-applying is Busted. Cool :thumbsup:

I suspected as much, from that vid from AMD where someone named "Castillo" (or something like that) from their server department showed how to do a cpu upgrade with just a bios upgrade (but for the sake of brevity, he just did a bios chip swap), then put in a new CPU without having to clean off old TIM from the heatsink and people were all hysterical in the youtube comments.

Thanks for the herculean effort and method :thumbsup:

 

[n7]

IDC, you've made a beyond fantastic thread completely saturated with fascinating and useful information

Thank you.

 

[Tikerz]

LOL, thumb print and clumsy hair. Love it!

 

[deimos3428]

I am surprised that the zero TIM results were so bad, and the thumb print and hair and toothpaste results were so not-so-bad. Thanks for all the wacky tests!

"Next, please try having a goat chew on the CPU for a while and then soak it in diesel fuel and set it ablaze. I've heard bulldozer posts much better stats here. NB: Keep the goat away from the flames, as this will invalidate the results."

 

[oceanside]

Awesome thread IDC, the best I've read in years!

 

[RussianSensation]

So IDC, what did you end up going with? Are you going to keep the H100 since you already bought the noise blocker fans?

BTW, this thread is too epic. So am posting to bump it to first page

 

[Jaeden]

Sexyyyyyyyyyyyy. This is like nerd porn.

you have no idea how much that line cracks me up!:biggrin:

 

[OVerLoRDI]

Again great work. Glad you did a test with Indigo Xtreme, I have been considering picking some up. Interesting about the metal film that it leaves behind. The idea of polishing your heatsink every time to remove the metal film is a bit of a deterrent though.

 

[Idontcare]

So IDC, what did you end up going with? Are you going to keep the H100 since you already bought the noise blocker fans?

BTW, this thread is too epic. So am posting to bump it to first page

Sorry it took me a month to reply, I stopped checking this thread for new posts a while back and somehow yours failed to register with me.

I'm keeping the H100 on my 2600K. Between the H100 and the noiseblockers it is a $250 investment.

The D14 is going on my HTPC which has a Q9650. Should be nice and quiet, and total overkill

 

[Rubycon]

I'm wondering if I should have my TIM (AC4) replaced with Indigo Extreme? The blocks will not be removed at all most likely. My load temps are in the upper 40s now. (X5690 at 4.50 / 1.375V)

 

[Elcs]

I'm wondering if I should have my TIM (AC4) replaced with Indigo Extreme? The blocks will not be removed at all most likely. My load temps are in the upper 40s now. (X5690 at 4.50 / 1.375V)

I'm slightly tempted too, mainly for the geeky/tweaking side of things.

I'm using the TIM that's pre-applied to the H100 and getting load temps of 62C after 2 hours of Prime95 (i2500k at 4.533 / 1.224v).

I think I might have a lot of headroom still. The above temps are with my Corsair fans running as slow as I can get them to run (below spin up voltages).

What about the CPU liquid metal thermal paste or metal wafers from Cool Laboratories?