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

[Idontcare]

But if you buy the H50/60/70/80/100 and proceed to replace its stock fans, then aren't you quickly eroding the value proposition? But then if you aren't aiming for top performance at low noise levels, what is the point of watercooling exactly? If you are only doing mild overclocks and want to spend $70-80, you can have a passively cooled HR-02 setup. I suppose if you absolutely dislike metal heatsinks and are aiming for aesthetics and the cool factor of watercooling, it's worth it. To me personally, the whole idea behind watercooling is better performance at the same noise level OR similar performance at a lower noise level that makes it superior to air cooling. Maybe I am underestimating how many people want the cleaner layout that Corsair's units offer.

Without a doubt it changes the value-proposition, but obviously the value proposition itself depends on how you are calculating the "value add" of the 3rd party cooling altogether.

If I am looking to keep my CPU at stock clocks then the value-proposition for just about any 3rd party HSF is going to be lacking.

If I am looking to hit 5GHz from my OC then the value-proposition of anything less-than-stellar cooling is going to be a waste of my money and time.

If I am looking to hit maximum clockspeeds while keeping the noise levels below some arbitrarily assigned threshold (e.g. 35db, 40db, 2π³ db, etc) then keeping the stock fans on most 3rd party HSF's is questionable.

Not everyone has the same goals and objectives in mind, that changes the context of the value proposition on an individual basis.

 

[RussianSensation]

IDC, no question, as I did mention, if you want the best performance, air cooling won't beat H100. If you want the look of a closed system, again, H100 is a great option.

I think H100 will shine especially once you start increasing the voltage on the CPU. I hope you have a great overclocking chip that gets to 4.8-5.0ghz so that we'll be able to see NH-D14 and H100 at their best

 

[Idontcare]

I'm questioning whether or not I should futz around with characterizing my setup prior to lapping.

There's no question I will lap the CPU and HSF's before I actually commit to one. Given the plethora of reviews out there already on the H100 and the D14, I'm leaning towards just getting on with it and jumping straight to the meat of the question I personally am most interested in answering - lapped surfaces, best performance at comparable noise levels.

I bought a noise meter as well as an IR gun. Got my multimeter setup on the probebelt. My Gskill ram is busted though, won't function at 2133 like advertised, will function at 1866, so I am in the process of RMA'ing the ram.

That will set me back a couple weeks easy. Good time as any to go ahead and lap the gear is what I am thinking.

 

[RussianSensation]

That will set me back a couple weeks easy. Good time as any to go ahead and lap the gear is what I am thinking.

I think the lapping will help both since neither of their base surfaces are polished to mirror-shine:

H100

NH-D14

 

[james1701]

Has anyone posted any reviews with a 980X or 990X on an H100. I am getting ready to start my rebuild next week, and I love how the H50 performed on my 980X with a couple of 1450 GT fans on it. I think this time around, I am going with the H100.

 

[Idontcare]

Has anyone posted any reviews with a 980X or 990X on an H100. I am getting ready to start my rebuild next week, and I love how the H50 performed on my 980X with a couple of 1450 GT fans on it. I think this time around, I am going with the H100.

I haven't seen a review like that, but just from futzing around with my H100 on a 2600K I would think anyone spending $1k on a 980X or 990X would want nothing less than an H100 to cool it IMO.

And I say that only in the sense that I am not all that impressed with the stock cooling capabilities of even the H100 on a mere 2600K. Coming from a vapoLS background, I expected water cooling (even the "toy" kind that is these turnkey corsair units) to be capable of keeping my stock-clock temps in the 40's. Instead I see 60's very easily.

 

[RussianSensation]

Has anyone posted any reviews with a 980X or 990X on an H100. I am getting ready to start my rebuild next week, and I love how the H50 performed on my 980X with a couple of 1450 GT fans on it. I think this time around, I am going with the H100.

http://www.xbitlabs.com/articles/coolers/display/corsair-h100_5.html

 

[james1701]

Thanks. That was fairly impressive with as much voltage as they were putting through the chip and still keeping out of the 80C range.

 

[steel108]

I am using the H100 on a 2600k OC to 4.6 @ 1.325 vCore. I didn't take pics of my 5 hour+ runs on Prime, but this should give you a general idea; I have 4 AP-15 in push/pull intake on the top of the CM 690 II Advanced:



My temps tend to peak after 15 minutes on Prime (did an 8 hour run without getting higher temps). The fans are set to 1200rpm via my fan controller; running at full blast dropped temps by a few C. Note: the supplied fans are LOUD.

Edit: in case you can see it, my high temps were:

49 C 56 C 52 C 52 C

I did my runs a few weeks ago, didn't take ambient temps, but it was warm (I live in LA and it gets pretty hot). Done at night without AC.

 

[Rifter]

And I say that only in the sense that I am not all that impressed with the stock cooling capabilities of even the H100 on a mere 2600K. Coming from a vapoLS background, I expected water cooling (even the "toy" kind that is these turnkey corsair units) to be capable of keeping my stock-clock temps in the 40's. Instead I see 60's very easily.

well i dont have the H100 i do have the H70 on my HTPC and i agree with you that i was dissapointed.

Which is why i still consider the "fake" watercooling setups a complete waste of time, either go all out real watercooling or stick with the top tier air coolers.

To be honest im super impressed with the NH-D14, im ramming 1.35V core and QPI into my i7 930@4.2 with HT which has to be drawing close to 250w at full load and the NH-D14 handls it like its a cakewalk, i love this thing.

 

[Idontcare]

Oh ho ho, holy Nikes! This is nice!

I just replaced the two stock Corsair "noizemakers 2000" fans on my H100 with four M12-PS Noiseblocker fans, PMW controlled and all fed into an Akasa Flexa FP5 5-way PMW splitter that is connected to the mobo's CPU fan header.

I set the fan profile in the bios to "silent" which basically means it keeps rpms really low until the CPU gets really hot, then it ramps fans quickly.

This system is dead-silent LinX @3.8GHz across all cores and my peak temps are in the mid-60's. I mean my laptop makes more noise than this thing (laptop has spindle drive, this is SSD).

I haven't done anything to tweak the system. Vcc seem rather high for a default clocked system at 1.360V. (its on auto)

Now, granted I just bolted ~$120 worth of fans to my ~$110 H100 cooler, so the sucker better deliver, but it is just so nice and quiet.

Corsair really ought to partner with a better fan OEM and replace their stock noizemakerz with something less 1990's and more post-2010.

Next up is the lapping

 

[BonzaiDuck]

Oh ho ho, holy Nikes! This is nice!

I just replaced the two stock Corsair "noizemakers 2000" fans on my H100 with four M12-PS Noiseblocker fans, PMW controlled and all fed into an Akasa Flexa FP5 5-way PMW splitter that is connected to the mobo's CPU fan header.

I set the fan profile in the bios to "silent" which basically means it keeps rpms really low until the CPU gets really hot, then it ramps fans quickly.

This system is dead-silent LinX @3.8GHz across all cores and my peak temps are in the mid-60's. I mean my laptop makes more noise than this thing (laptop has spindle drive, this is SSD).

I haven't done anything to tweak the system. Vcc seem rather high for a default clocked system at 1.360V. (its on auto)

Now, granted I just bolted ~$120 worth of fans to my ~$110 H100 cooler, so the sucker better deliver, but it is just so nice and quiet.

Corsair really ought to partner with a better fan OEM and replace their stock noizemakerz with something less 1990's and more post-2010.

Next up is the lapping

Do it while watching South Park or Looney Toons. The hour spent for each surface won't seem like a tedious waste of time, and you'll earn yourself a Centigrade advantage of some sort.

 

[Spikesoldier]

you gonna lap cpu and peltier idc?

 

[Idontcare]

you gonna lap cpu and peltier idc?

Yeah, both, but its not a peltier, just an H100.

I did lap my vapophase LS though, so lapping the sub-zero stuff is on my resume

I loves me some South Park, but I got kids now so it'll be Dinosaur Train or Word Girl for me.

 

[RussianSensation]

What happened to your Vapor systems? I guess they no longer sell brackets for newer sockets or you ended up selling them?

 

[james1701]

Speaking of swapping fans. Will either the Haf X or 800D have enough room in the top to run a push pull with some Scythe GT fans? I had the 1450rpm models on my H50, and they did a smash up job for how quiet they were. I might try the 1850rpm's on the H100 when I get it.

 

[Idontcare]

What happened to your Vapor systems? I guess they no longer sell brackets for newer sockets or you ended up selling them?

Sadly my once proud, $1k investment, VapoLS gear sits in my basement, all mothballed up for use "some day".

It is true that I have not seen clamshell kits to make it work with anything modern, socket-wise, but I worked hands-and-knuckles on that thing for years to the point that I'm fairly confident now I could fashion my own ghetto-styled clamshell contraption to get it to work with today's sockets just the same.

But the noise, its the noise that really got to me. Just wasn't worth the extra 20% overclock that the -40C afforded over room-temp cooling.

I still have grand visions of dragging that thing out again, some day, and ripping up some benches, but not today

 

[Idontcare]

I think the lapping will help both since neither of their base surfaces are polished to mirror-shine:

My ram is headed off to GSkill today for rma, so now was the time to begin the lapping.

As Russian noted, the NH-D14 surface is not at all polished. It does appear as if it were machined (milled) to be flat by-eye though:

Before I started sanding, I added a cross-hatch pattern with a red sharpie pen.

I started with 220 grit, took a few passes (maybe 10) and then inspected to see how even or uneven the surface really was.

Ouch, not good. My NH-D14 had a high-spot right in the center, as well as a high edge to the left.

After a few more passes on the 220 grit (maybe another 10 passes) I inspected it again:

Ugh, double not-good. I have fully removed the original high spots but now we see I also had a low trough too (the remaining gray island to the right)

The low spot was actually really really low, I must have spent nearly 10 minutes sanding on the 220 grit to get the trough down to this barely noticable (but still there) small spot on the right:

Finally, the spot was gone and the HSF surface was flat after about 15 minutes on 220 grit (which is a really long time to spend at the coarse of a grit).

Moved on to 400 grit, here's the results after 400:

After 1000 grit:

And after 3000 grit:

Looks pretty flat to me

I haven't got to the 2600K or the H100 yet, but I did get my Q9505 lapped (going into a HTPC with either the NH-D14 or the H100 when I decide which one goes with the 2600K at the end of all this).

 

[Rifter]

My ram is headed off to GSkill today for rma, so now was the time to begin the lapping.

As Russian noted, the NH-D14 surface is not at all polished. It does appear as if it were machined (milled) to be flat by-eye though:

Thats actually pretty opposite my experiance with the D14, mine came almost completly perfectly flat and pretty much only needed a polish i hardly removed any material. Guess its luck of the draw like anything else though.

My i7 930 on the other hand had a huge depression in the heat spreader and took me many hours to get down to a flat and polished state.

 

[beginner99]

how important is this flattening and polishing?

I just unpacked my stuff and installed. Might explain why my temps are higher than all reviews show?

 

[Rifter]

how important is this flattening and polishing?

I just unpacked my stuff and installed. Might explain why my temps are higher than all reviews show?

Generally lapping is considered extream, only really advanced overclockers do it or those who are really anal about temps. I have seen it make 10-15c differences in the past over stock but you must also take into consideration that its not only the lapping making that difference as i always replace the TIM with quality stuff and make sure the heatsink is properly seated which it may or may not have been from the factory, i have seen a few GPU's with iffy heatsink mounting from the factory.

 

[Idontcare]

how important is this flattening and polishing?

I just unpacked my stuff and installed. Might explain why my temps are higher than all reviews show?

It helps, never hurts, to lap the surfaces. Assuming you lap them correctly of course.

It's become old hat for me though, I've probably polished around 10 cpu's by now, plus 10 HSF's.

Compared to how tedious and anal I made the process the first time, out of unfounded fears, the process is rather quick and easy now.

Temps can be off for any number of reasons. Your ambient temps may be different, you may be testing inside a case but the reviewers tested in the open, your specific CPU may simply operate at different Vcc's for any given clockspeed and thus be using different amounts of power and heat.

And you may need to re-seat your HSF, use a different TIM, use more or less TIM, etc etc.

Generally speaking, a good indicator that you have flatness issues and you need to lap is the difference in temps across the cores. My 2600K and H100 setup has about an 8C delta between Core 1 and Core 3.

I expect this will decrease to less than a 2C delta once I've lapped them (going by experience), the hotter core will decline the most, the coolest core might only decrease in temp by 1-2C. But the distribution will tighten up and the move to slightly cooler temps on average.

I have had one CPU though where lapping simply made no difference, the issues were under the IHS itself - either poor contact between the silicon and the IHS itself or the core on the silicon was just different than all the rest. As they say, you can't fix stupid, and some cores just can't be helped when it comes to their temp issues.

 

[Idontcare]

Thats actually pretty opposite my experiance with the D14, mine came almost completly perfectly flat and pretty much only needed a polish i hardly removed any material. Guess its luck of the draw like anything else though.

My i7 930 on the other hand had a huge depression in the heat spreader and took me many hours to get down to a flat and polished state.

Yeah, that's the take home message, even with today's top of the line HSF's there is no such thing as being allowed to assume the surface is flat just because someone else found their's to be flat.

Honestly I was quite surprised at how warped the surface of my D14 turned out to be, it had such a meticulously milled surface I was convinced I was setting out to waste my time.

The Q9505 on the other hand, I did all the same spot-checks with the sharpie hatch-pattern and inspections but I didn't bother taking pics because it was near perfectly flat from the start.

 

[stahlhart]

Looks pretty flat to me

Now I want to lap my SA. NICE work....

 

[RussianSensation]

Sexyyyyyyyyyyyy. This is like nerd porn.