The 10,600-degree heat feels a bit too hot for my liking.
The 10,600-degree heat feels a bit too hot for my liking.
Hey everyone, I'm using a 10600K right now and getting worried about how hot it gets. I want to tell you all what I know here. I have a light overclock on the CPU where every core is boosted up by 4.8GHz. The ring multiplier is set to 44x, so no AVX offset is running either. In the BIOS, voltage is 1.280V, but under load it drops slightly to around 1.260V because of how the memory controller works.
For the cooler, I'm using a Noctua D15-S, and right now I only have one fan spinning on the right side. My motherboard is an MSI Z490 Tomahawk, and my case has three fans: two in front for air intake and one at the back to blow hot air out.
I am wondering if I've hit the limit of this cooler or if I should just take it apart and re-paste everything? I'm leaning towards changing things because when I built this PC last week, the temps got way higher than they are now, and I knew something was wrong. Turns out, the screw that holds the heat sink in had only about 3/4 of a turn left to go before it stopped turning. After I tightened it all the way down, the temperatures dropped a lot. But then the computer ran fine for two days or so until I realized my mistake and checked it again.
I'm wondering if that uneven pressure caused some air gaps between the cold plate and the CPU chip (the IHS), which messed up my thermal paste job? Or is it just because of how well the cooling solution is applied that those temps are higher than usual?
Here are the numbers: The room temperature is about 22 to 23 degrees Celsius. When the computer is idle, the temp goes from 35 to 38C. During Prime95 26.6 tests, it hits 85 to 88C. Cinebench R20 comes in at 78 to 82C. AIDA64 shows 65 to 68C even when the game is running Battlefield 5 Multiplayer, which I think uses my CPU the most. Other single-core games that aren't super intense keep the temps in their high fifties.
When Prime95 runs under load, the package power usage is around 140W. This cooler is definitely rated for more than that; it can handle at least 230W easily. Before, I thought about trying a 5.0GHz all-core overclock with less than 1.35V of voltage because sometimes luck on the lottery helps me get lower voltages. But right now I don't think there is any room for more speed or lower voltage.
Even though some reviewers said they didn't recommend going over 4.8GHz on air with this chip, I was skeptical when I heard it first and still believe that's okay sometimes. Either way, I'm taking the cooler off tomorrow to put a brand new layer of thermal paste back on just to satisfy my curiosity. If anyone else has been running a 10600K on air and is overclocking, please tell me what their temperatures are in things like Prime95, Cinebench, Blender, or even games like Battlefield since those show how much power the CPU uses.
I know that for my normal gaming use, I don't see any signs of damage to the CPU, so everything is safe right now. But just wondering if there is more headroom with this cooling setup.
Yep, you are in the right area now. Adding an AIO will help a bit and if you can tweak it further, that's good too. The 10600K has lots of room to go even at just 1.3v. My 10700K hits up to 5.1GHz at 1.295v with Turbo LLC and the 360mm AIO is fine for temperatures, which means your current setup with good cooling could be pushed further. An all-core 5GHz with a 280mm AIO should work well even without an AVX offset, keeping temps under control.
1) Put the cooler back on... I don't know why you didn't turn everything off and fix it right there when you found out it wasn't safe to touch. Since it wasn't locked down properly, moving it from side to side could easily make things fall over. 2) You don't need an AVX offset? Yeah, if you aren't doing that, you won't be able to get your 5.0ghz overclock on that cooler at all. At the same frequency as SSE, AVX is actually faster, but it needs more voltage to stay stable, which means using more power and making even more heat. Some apps - not many right now - run them, especially things like Cinebench R20...
No, I'm pretty sure I need a -1 or -2 AVX offset at 5.0ghz. Right now, my CPU isn't running one on the 4.8ghz mode. And you're right, I should have just shut it down and re-put the cooler back together immediately when I noticed that weird screw issue with a loose thread on the post. I didn't do that because it was late at night, temps got okay enough, so I thought everything worked fine (which is my bad). Since then, I've been thinking about how stupid of a decision I made lol. I haven't got time to put it together yet. I spent most of today building another PC for a friend but am still planning on doing it tonight. I'll tell you all when I get back. The cooler didn't move much at all. That top screw was stuck right where it wanted to go. The bottom one was 90% tight and then suddenly gave up completely after a third turn. My guess is that the uneven pressure between the cold plate and my heatsink might have messed up getting good paste coverage. It's the only thing I can think of.
The re-mount has definitely gotten cooler again. When I first put the cooler in, I used the Noctua NT-H1 paste that came with it. For this run, I grabbed some brand new AS5 from my parts order and put a little bit of it on top of the heatsink in a straight line. To be fair, I only tried the AS5 to see if something different would help, not because I hated the old Noctua paste before. Neither is better than the other, or anything like that. Both idle and load temps have improved. Back then, my idle temps were 35 to 38C, now they are 32 to 34C. When running Cinebench R20 and Prime95, I saw about a 4 to 6 degree drop in heat during those tests. I never went over 80C while running R20 before; the temps stayed steady around 74 to 76C, with just one core hitting a peak of 78C. In Prime95, the small FFT on one core reached 82C, but all my other cores stayed at 78 to 80C. Those numbers feel much more comfortable for me. This is true because there was an average room temperature of about 23 degrees Celsius back then, and I was just using normal fan settings without pushing them hard first. It's hard to judge how good the thermal paste really is once you take the cooler off; maybe I used a little too much on my first try, or maybe it spread unevenly because of pressure issues. But being able to see an improvement is still great for me.
Big pastes are about a centimeter away. That feels weird calling them "high end." I think it's probably just about getting that screw mount right.