[SOLVED] 10600K temps a little concerning.

[RetroCNY]

Hi all,

I'm currently running a 10600K, and I'm a little concerned about the temps I'm seeing. I'll try to provide as much information as I can here.

I'm running a mild 4.8ghz all core overclock on the 10600K. Ring multiplier is 44x. Not running an AVX offset. Voltage in bios is set to 1.280 but with LLC settings, actual vcore under load is about 1.260. As far as the cooler, I'm running a Noctua D15-S. Only using the one fan for right now. Motherboard is an MSI Z490 Tomahawk. Case is a Meshify C with 3x 120mm fans. Two in the front, one for rear exhaust. I'm wondering if I've just reached the limits of this cooler(which I find a little hard to believe) or if I should go ahead and re-paste and remount the cooler. I'm leaning towards the second, but I'd love to hear from other 10600K users that are overclocking on air, to see what kind of temps they're seeing in applications like Prime95 26.6, Cinebench R20, Blender, etc. The reason I'm leaning towards the second option is that when I initially built this pc last week, I noticed temps that were even higher than what I'm currently seeing and I knew something had to be wrong. It turned out, the bottom mounting screw on the heat-sink had about 3/4 of a turn left to go before it bottomed out, and I must have had an air-gap between the cold-plate and IHS because when I tightened the screw until it bottomed out, I saw temps improve. However the computer ran like that for 2 days or so before I recognized the problem and I'm wondering if that slightly uneven pressure may have screwed up my thermal paste application by affecting the overall coverage on the IHS...and the temps I'm still seeing could be a byproduct of that. Or, if I really am just thermally limited by my cooling solution.

Here are some temps:

Ambient room: 22-23c.
Idle: 35-38c
Prime95 26.6: 85-88C
Cinebench R20: 78-82C
AIDA64: 65-68C
Battlefield 5 Multiplayer(Probably the most CPU intensive game I have installed): 68-70C
Other single-threaded games that aren't as intensive: High 50s.

My CPU package power under P95 26.6 load is 140w. This cooler is definitely rated for way more than that, I believe 230w at least. Originally, I was shooting for a 5.0ghz all core overclock with 1.35 volts or less depending on my luck with the lottery, but currently I just don't think I have any headroom to do it thus why I'm running 4.8 at the moment. I have seen other reviewers of the 10600K claim that running past 4.8 on air with this chip wasn't recommended and maybe they're right, but I was skeptical of that when I heard it initially and still sort of am.

Either way, I'm going to pull the cooler off tomorrow, and remount it with fresh paste just to sate my curiosity here. If anyone is running a 10600K on air, I'd love to hear your overclocking results and temperatures. I know the temps I'm seeing under my typical use-case, which is gaming, are well within safe limits and I don't have any concerns of potential damage to the cpu. Just wondering whether I should be seeing more headroom with my cooling solution.

 

[Phaaze88]

1)Remount the cooler... IDK why you didn't shut everything down and redo it then and there when you discovered that it wasn't fully secure. The cooler could have easily shifted moving from the horizontal to vertical position since it wasn't completely secured.

2)No AVX offset, eh? You're not going to be able to get your 5.0ghz OC on that cooler if you don't.
At the same frequency as SSE, AVX is faster, but requires more voltage for stability - but that also means more power drawn and more heat produced.
Some applications - not many at the moment - do run them, especially Cinebench R20...

 

[RetroCNY]

1)Remount the cooler... IDK why you didn't shut everything down and redo it then and there when you discovered that it wasn't fully secure. The cooler could have easily shifted moving from the horizontal to vertical position since it wasn't completely secured.

2)No AVX offset, eh? You're not going to be able to get your 5.0ghz OC on that cooler if you don't.
At the same frequency as SSE, AVX is faster, but requires more voltage for stability - but that also means more power drawn and more heat produced.
Some applications - not many at the moment - do run them, especially Cinebench R20...

No, I'm definitely under the impression that I'd need a -1 or -2 AVX offset at 5 .0ghz. I'm just not currently running one at 4.8 all core. And you're right, I should've just shut it down and remounted as soon as I noticed the issue where one of the mounting screws had a little more room to thread onto the post. I didn't because well, it was late at night, and the temps improved to reasonable levels so I just assumed(which is my fault) that everything was OK. Since then, I've begun questioning my decision making process lol. I haven't had a chance to re-mount yet. I spent most of my free-time today building another pc for a relative but I'm still planning on doing it tonight. I'll report back once I do.

The cooler didn't shift as far as I could tell. The top mounting screw was totally bottomed out, couldn't go anymore. The bottom mounting screw was say 90% tight, and 3/4 of a turn bottomed it out completely. My guess is that that bit of uneven pressure between the cold plate and ihs possibly may have interfered with getting good paste coverage. It's the only thing I've been able to think of.

 

[Phaaze88]

All good then. Keep us posted.

 

[RetroCNY]

Alright, the re-mount has definitely improved temperatures once again. When I initially mounted the cooler, I used the supplied Noctua NT-H1 paste that came with. This time, I decided to use some brand new AS5 I had ordered when picking out the parts for this build. I used about a rice grain's worth in a vertical line on the IHS. To be fair, I only used the AS5 just to try something different, not because I didn't like the Noctua NT-H1 that was previously on the cpu. Not claiming one is better than the other, or anything like that.

Idle and load temperatures have both improved. I was originally idling around 35-38C, I'm now idling around 32-34C. Load temperatures in Cinebench R20 and Prime95 26.6 have seen roughly 4 to 6C worth of improvement. I'm no longer exceeding 80C during R20 runs, temperatures hold steady in the 74-76C range, with a one core peak of 78C. Prime95 26.6 Small FFT peak temperature on one core is now 82C, where as all the remaining cores are hovering around 78-80C. Much more comfortable with those numbers. This is with a 23c ambient room temperature, and I was just running normal fan curves for both my CPU and case fans. I didn't max them out prior to starting the test.

It's hard to tell when looking at thermal paste once you've removed the cooler, it's possible I may have used slightly too much on my initial application. And it looked like I could've had better spread on the lower portion of the IHS, as was my theory due to the mistake I made/uneven pressure, etc. Happy to be seeing an improvement though.

 

[Phaaze88]

The high end pastes are like 1C apart - if that.
Feels weird applying the term high end to bloody thermal paste...

I'm sure fixing the mount is what really did the trick though.

 

[RetroCNY]

Yeah, it had to of been. I've spent the last hour or so getting 5.0ghz all core, -1 offset, @ 1.34 vcore seemingly stable. I say stable loosely because 1hr is not enough time to know for certain, but no WHEA errors in HWINFO or BSODs. Assuming the cooler is now working as it should, this definitely feels like the limit for this cooler @ this voltage range based on the temps I'm seeing now. P95 exceeds 90C, peak of 93C but typically 89-91C. Cinebench is 85-87C, blender bmw benchmark is about the same, and a couple rounds of battlefield 5 multiplayer saw temps hovering right around 75C. BF5 really is an intensive game and it utilizes AVX, so not all that surprising. A little warm for gaming temps as far as my tastes go, but safe regardless.

This seems fairly typical based on some of the reviews I've read. Tom's Hardware review of the 10600K at 5.0ghz on a 280mm H115i were seeing temps in the same ballpark compared to what I'm seeing, albeit they left LLC on auto which gave them a higher 1.368 volts under load compared to my 1.340 volt @ load. I'm slightly warmer than them, maybe 3-4 degrees max based on what they're reporting, but that's not too terribly surprising. I am planning on switching to an AIO in the future, all the 280MM AIOs I was interested in were unavailable when I was parting this build out, so I just went air for now for the availability. In the mean-time, I may grab a 2nd fan for this cooler just to see if I can get another 1 or 2c worth of cooling out of it. I do love this cooler though, compatibility is a non-issue. Not disappointed truthfully with the results though. 5.0 ghz on this chip with air cooling seems more than acceptable.

 

[vMax]

Yep you are in the right ball park now. An AIO will make a bit of a difference and if you can fine tune further maybe a little more. The 10600K has a lot of headroom that can be unlocked even at just 1.3v. My 10700K will hit 5.1GHz at 1.295v, Turbo LLC and with a 360mm AIO is more than okay on the temps front which shows that your setup with the right cooling can and should be pushed further. An all core 5GHz with a 280mm AIO should be more than achievable with temps well under control even without an AVX offset....