6700k Thermal Paste Application & Expected Temps

[Ghoststalker]

Hey folks,
I bought some Noctua NTh1 for my new 6700k. I am using an H100i v2 with it. The instructions say to place a 4-5 mm dab in the center and place the heatsink on, twist it, and you're good to go. I noticed this does not cover the entire heatspreader which didn't appear right to me.

What temperatures should I expect at load if the ambient temperature is around 79F? Last night at stock clocks in The Division the Package Die was peaking at 64C, which is quite a bit higher than my 2600k @ 4.5ghz was (maxed at 58).

 

[primetime]

i would try to redo it.......that is a tad high for just gaming, People can talk bubbles all they want idk cause i cover the hole surface that makes contact. you would think water cooling would works better than that. What kind of temp are you getting with real bench at 100% or whatever stress test?

 

[Araxie]

very vague information and even more the comparison as we don't know how high OC'd is the 6700K or if its at stock...

Is normal to find a huge difference between the i7 2600K and the i7 6700K starting with the first one used soldered lid and the later just a crap TIM, this change entirely how the dissipation works from the CPU die to the internal heatsink and from this to the cooler... Of course this is just guessing both chips are overclocked and tested under same circumstances, is also worth to mention that the H100i v2 utilize variable PUMP speed based in temp while the old H100 was a fixed high RPM, what settings are you using to control the h100i v2? quiet profile for fans and pump? also worth to mention that as you may know coming from a 2600K some programs and temperature reading apps do not use real package die temp (Aida64 do this) but an average of all the core temps so pay attention as you must know to each core temperature that's what's truly important..

And last but not less important with years of experience using Noctua NT-H1 (used to be a truly fanboy of this TIM) never do that spread method, best spread method for that paste is just to put a very small amount of TIM over the cpu and spread yourself manually until reach a very thin and even layer that's how it work better for that specific TIM.

 

[Ghoststalker]

Sorry was at work and didn't have time to provide details. It is running at stock. Looking more for tips on how to use this thermal paste best, thanks for the guidance. I'm going to run some prime95 now and see how my temps end up. I may redo thermal paste tomorrow.

 

[primetime]

And last but not less important with years of experience using Noctua NT-H1 (used to be a truly fanboy of this TIM) never do that spread method, best spread method for that paste is just to put a very small amount of TIM over the cpu and spread yourself manually until reach a very thin and even layer that's how it work better for that specific TIM.

This is pretty much how i apply it and people swear it creates air bubbles. lol its always worked well for me

 

[mothman]

It's not the Noctua tim it's the install of your H100i causing your high temps. The notoriously poor mounting system of the H100 or any other Corsair AIO makes good contact virtually impossible. Ditch that H100 and get any Noctua cooler.

 

[Ghoststalker]

It's not the Noctua tim it's the install of your H100i causing your high temps. The notoriously poor mounting system of the H100 or any other Corsair AIO makes good contact virtually impossible. Ditch that H100 and get any Noctua cooler.

First I've heard of that and I was using an h100i (original) for quite a while. This one also had very good temps once I had an issue and I got his in the RMA. I'll try reapplying the thermal paste tomorrow and see how it goes.

 

[Araxie]

It's not the Noctua tim it's the install of your H100i causing your high temps. The notoriously poor mounting system of the H100 or any other Corsair AIO makes good contact virtually impossible. Ditch that H100 and get any Noctua cooler.

This is baseless statement.. 90% of the issues with corsair h100 units (CooliT models) are due user error in the installation, the backplate have a unique mounting orientation and it's that what cause the cooler to not fit properly. Other 5% of the issues are related to some motherboards that utilize a very thin PCB which cause the cooler to not fit properly and be loose requiring to aid it with washers or o'rings... and we'll there's of course another percent of mechanical failures.. but that's far of "notorious poor mounting system" which its also far away different from asetek made units which utilize a different mounting system and also its different from newer editions of the cooliT mounting system like the used in the h100i v2. Totally different from the original h100i and h100...

However yes, you are pointing in a posible right direction which may be bad mounting but at the same time are certainly giving a poor advice saying him to buy a new cooler.

 

[primetime]

This is baseless statement.. 90% of the issues with corsair h100 units (CooliT models) are due user error in the installation, the backplate have a unique mounting orientation and it's that what cause the cooler to not fit properly. Other 5% of the issues are related to some motherboards that utilize a very thin PCB which cause the cooler to not fit properly and be loose requiring to aid it with washers or o'rings... and we'll there's of course another percent of mechanical failures.. but that's far of "notorious poor mounting system" which its also far away different from asetek made units which utilize a different mounting system and also its different from newer editions of the cooliT mounting system like the used in the h100i v2. Totally different from the original h100i and h100...

However yes, you are pointing in a possible right direction which may be bad mounting but at the same time are certainly giving a poor advice saying him to buy a new cooler.

yea i also agree...just put a little more focus in making sure its mounted as "even and snug" as possible. Use the method Araxi described with the paste. You don't want any bare areas that make contact like before. Make sure the pump and fan speed are high enough to do the job. Noise doesn't bother me much so i might at least try max speed. I predict a 10 degree drop in temps tomorrow

 

[Ghoststalker]

Went ahead and reapplied the TIM tonight since I was bored. I did it this time using the Noctua recommended install (drop and twist HSF on). Running prime 95 now, looks like "package temp" maxes at around 77c, but that is very briefly and it's usually around 67. The individual cores vary from 58-77 based on their load.

As for the h100i v2 mounting, it doesn't have the issue that the older ones do as it's the asetek model. It seems to be fairly well snug, I can't tighten the screws any further (by hand or with screw driver).

The temperature liquid is hovering at around 34c (this is after only 10 minutes) so I'm wondering how good the contact is. I know on my 2600k it would cling to around 40c when I was gaming. I'm using "balanced" settings in Corsair Link. I read some folks mentioned pushing down on the coldplate to see if you notice a temp drop on the CPU, which would indicate a bad contact. Do you guys think that is worth doing?

I am going to try on overclocking tomorrow and see where I end up.

 

[primetime]

Went ahead and reapplied the TIM tonight since I was bored. I did it this time using the Noctua recommended install (drop and twist HSF on). Running prime 95 now, looks like "package temp" maxes at around 77c, but that is very briefly and it's usually around 67. The individual cores vary from 58-77 based on their load.

As for the h100i v2 mounting, it doesn't have the issue that the older ones do as it's the asetek model. It seems to be fairly well snug, I can't tighten the screws any further (by hand or with screw driver).

The temperature liquid is hovering at around 34c (this is after only 10 minutes) so I'm wondering how good the contact is using "balanced" settings in Corsair Link. I read some folks mentioned pushing down on the coldplate to see if you notice a temp drop on the CPU, which would indicate a bad contact. Do you guys think that is worth doing?

I am going to try on overclocking tomorrow and see where I end up.

i cant see any reason not to try doing that.......keep a close eye on temps and see what happens. yea i would. Set it on max speed as well so you can compare it to what reviews got at least for a bit to see the difference....might even consider push pulling the fans at some point as well just as a test

 

[Ghoststalker]

i cant see any reason not to try doing that.......keep a close eye on temps and see what happens. yea i would. Set it on max speed as well so you can compare it to what reviews got at least for a bit to see the difference....might even consider push pulling the fans at some point as well just as a test

Thanks. See attached for how much thermal paste I used this time. Its a fairly thin amount on the top, the pic angle makes it look bigger than it was. The H review on this paste showed this method was a bit better.

 

[primetime]

long as it looks like this after removing id say you be ok

 

[Ghoststalker]

I didn't want to remove it afterwards due to concern about gaps, however temperatures are looking decent so far. Around 70c. I am switching away from p95 for stress testing after doing more research, using an x264 stresstest now. I was able to boot into windows at 4.6 with adaptive voltage set to 1.35, but instant crash in the stress test. Tried upping voltage a bit but no dice. Going to keep playing with it, testing at 4.4 now before I work any higher.

Downside is I can't find an LLC/VDroop setting on this motherboard, which kind of sucks. If I can get back to 4.5 (I think that should be easy enough) I think I'll be happy for now, but may upgrade mobo down the line.

 

[Trimlock]

Oh yea, don't use prime to stress test, you'll get some stupid temps with it.

When I get a new cooler I like to do a test run. Put it together, except for the case, run a stress test and then remove the heat sink to ensure I have a good spread of the thermal paste (like in primetime's photo). If all is good then it goes into a case.

 

[Ghoststalker]

Any sense of what load temperatures I should expect using something like x264? I was around 69-70c on the package (1.33v actual according to HWInfo). Seems good enough.

 

[primetime]

i would try and duplicate how [H] tested and compare

The mention using AIDA64 Extreme Edition. It's hard to say how it compares to what your doing