[SOLVED] i7-12700k with H115i hitting 90C in Cinebench ?

[Neptune77]

Just finished a new build with an i7-12700k being cooled with a Corsair H115i as a front mounted exhaust in a Lian Li O11 Air Mini. Also, I installed it with Noctua NF-P14s redux-1500rpm fans instead of the stock ones. My CPU temps are higher than I expected. When it comes to AIOs, I have never used anything other than a single 120mm AIO paired with an i7-4790k on my previous build, so I didn't really know what to expect from a 280mm rad, but I definitely didn't think they would be getting as high as they are.

I am getting max temps between 88-91C when I run Cinebench and I am wondering if something is wrong. This is after remounting the pump (before they were getting up to 97C), and I am nearly certain that at this point the mounting pressure is correct, and that it is making good contact with the CPU. Also my video card also has a 240mm AIO that I have as a top mounted exhaust, so there really shouldn't be too much heat inside the case, and I have 5 intake fans.

Shouldn't I be getting better temps with an i7-12700k paried with 280mm rad? I'm attaching a picture of my setup so you can get a better idea of the situation.

https://imgur.com/pEnUqAC

 

[Phaaze88]

a Corsair H115i as a front mounted exhaust

Why? Look at your own picture.
Bottom intake typically doesn't bring in that much air compared to other areas, and what does make it through is sucked in by the gpu's fan, because it's closer.

Side intake, look how much potential surface area is impeded by the radiator, gpu and cables; can barely see the lower fan, and about half the other one is visible.

The rear intake, is pretty much exclusive to the top radiator, but there's something else - it's pretty minor, but still: the hybrid gpu is exhausting out the top, but also out the rear... and because heat that isn't under the influence of a fan rises, that air gets pulled right back in by the rear intake and yanked through the top radiator.

Other concerns:
-You don't run benchmarks like these on auto voltage, or at the minimum, don't leave Vcore on auto. It'll always be sub-optimal.
-Motherboard Enhanced Turbo feature was enabled or not. Some Z boards have it enabled by default or as part of the optimized defaults. It's an auto overclock on auto Vcore . Best left disabled. Asus calls their version Multi Core Enhancement.



TL;DR: I have concerns that you may not be providing the cpu AIO enough air, specifically under heavy load, running with Vcore on auto, plus the possibility of MCE enabled.

 

[Neptune77]

This is 100% not an airflow issue. I ran Cinebench again with all the panels off, and the temperatures didn't change even a little bit. As for Vcore and MCE, they are both on auto, but shouldn't I be able to have those turned on and still get lower temps with a 280mm rad? This CPU doesn't run that hot does it?

 

[Nine Layer Nige]

The whole point of running a benchmark is to stress the cpu to its max performance, which means things are going to get very hot.
How does your Cinebench score compare to others ?

 

[Phaaze88]

You took a desk or box fan and used it to blast air inside the PC, right?
Simply taking the panels off what should be an already airy chassis - did doing this really increase the air volume moving through that front radiator? My concern with airflow was that space behind(?) the radiator.

As for Vcore and MCE, they are both on auto, but shouldn't I be able to have those turned on and still get lower temps with a 280mm rad. This CPU doesn't run that hot does it?

No, those settings can easily overwhelm ANY cooler, and yes it can run that hot, as you are now seeing. You missed quite a bit jumping from Devil's Canyon to Alder Lake.
They became faster(DUH) yet hotter.
They're practically maxed out from the start, unlike the older cpus, which had quite a bit of headroom left. People were pushing 1ghz OCs on old Sandy Bridge with ease... would need some very exotic cooling to do the same with these.
They became harder to cool with each process node shrink = greater thermal density in the package, and changed from paste under the heatspreader, to solder, to shrinking the height of the die and raising the heatspreader's, to try and deal with that.

More voltage than the cpu needs + an OC will make cooling it more difficult.
The excess voltage is done because silicon quality can't be predicted, so the vendors set the boards like that to account for the worst bins as a guarantee the cpu runs stable as advertised.
The use of enhanced turbo is more of a measuring contest between them.
You as the end user tend to get bitten by these settings more often than not.

 

[Neptune77]

Of course not. I was trying to replicate an open air test bench scenario, and there was already a ton of extra airflow from the case fans. I'm surprised anyone would even try suggesting that there would still be an airflow issue after removing the panels. I can assure you that if you had been here in person for the test, you wouldn't be thinking that. I thought that you should be able to stay relatively cool with a light OC and a 280mm rad though, have you done any testing or research on this particular chip yourself? From what I was seeing online, my temps are higher than usual. I see that Nine Layer Nige is running a 12th gen intel as well, although the 12600k. I'm curious what kind of temps you get with Cinebench. Here is a screenshot of my score since you were interested. I'm wondering what other people who have similar setups (same chip, similar cooler) are getting temp wise with Cinebench as well. If anyone in that category sees this, you should chime in.

https://imgur.com/a/2Ne7kkB

 

[Nine Layer Nige]

CineBench R23 on my i5-12600k stock, aircooled = 17193
WHInfo results
P-Cores max = 4889mhz
E-cores max = 3392mhz
Temps as follows:
P-core 0= 75c
P-core 1= 72c
P-core 2= 77c
P-core 3= 77c
P-core 4= 71c
P-core 5= 78c
E-core 6= 57c
E-core 7= 57c
E-core 8= 57c
E-core 9= 57c

You might want to look at the GamersNexus review video Steve did on the i7-12700K. He talks about thermals and a full set of benchmarks

View: https://www.youtube.com/watch?v=B14h25fKMpY

Also, Hardware Unboxed review here 13 minutes in ,temps

View: https://www.youtube.com/watch?v=vlngwUuDYoc

 

[jnxzi94]

Maybe the cooler is badly connected to the CPU or there is an issue with thermal paste.
Also, what I noticed is that your tubes are warped all over the place, does it make a crackling noise?
I have a i9 10850K @ 5 GHz and Kraken X63, max. temp on full load is 60 C.

 

[Nine Layer Nige]

Ahh, ive taken another look at the photo and see that the pump tubes are at the bottom.
Should the pump tubes be at the side, as trapped air might accumilate at the top of the block and may affect cooling ?
Worth reading this thread...

Question - AIO Orientation right or wrong? How screwed am I?

The GamersNexus video has me and a lot of other people worried. I was just wondering, how's this orientation? Not sure if I can flip it the other way, and I'd rather not if I can help it. If I understand it right, if the radiator is higher than the pump block I won't run into any issues, I...

forums.tomshardware.com

 

[jnxzi94]

Ahh, ive taken another look at the photo and see that the pump tubes are at the bottom.
Should the pump tubes be at the side, as trapped air might accumilate at the top of the block and may affect cooling ?

It's good that the tubes are coming from the bottom of the radiator, but they should be corrected so they make more of a straight line as we speak. You could try to put them from the radiator, coming from behind the GPU, so they don't make a knot.

Also, I would reapply thermal paste and remount the pump so it's not causing any issues.

Just follow this example:
https://hardforum.b-cdn.net/data/attachment-files/2021/01/420427_IMG_0236.jpeg

 

[Nine Layer Nige]

Yes I agree with the rad tubes at the bottom is correct, but I am specifically pointing to the orientation of the pump block where the tubes come out the bottom. Shouldnt they come out the right hand side ?

 

[jnxzi94]

As to mounting, I think the normal orientation that sends air out the rear exhaust is best.

Yes I agree with the rad tubes at the bottom is correct, but I am specifically pointing to the orientation of the pump block where the tubes come out the bottom. Shouldnt they come out the right hand side ?

Yes, they should come from the right side, because now it's backwards, as you can see from the logo of the RGB.

 

[Neptune77]

I have already remounted and repasted. That improved the temps to what I am currently getting, as mentioned at the beginning of this post. As for the orientation of the pump, I originally had it mounted with the tubes coming off the right, and the logo right side up. I had a lot of trouble getting the video card in though, so I ended up changing it to what it is now.

Neither of the tubes are taking any extreme angles, so I kind of assumed that it would be fine. It isn't making any weird noises, to answer your question. Is there some way I can tell if this is causing a problem by looking at coolant temps? How much should the coolant temps normally increase during a standard 10 minute Cinebench run? As long as the coolant temps are behaving normally, shouldn't that mean that the pump tubes are fine as they are?

 

[Nine Layer Nige]

I think the concern that I have read and watched is that air may get trapped at the top of the pump if the tubes are at the bottom. Trapped air means the coolant is not touching the entire coldplate so could impact cooling efficiency of the CPU.
By having the tubes coming out the right, any trapped air can escape.
You could test this by remounting the block with tubes right side but to do so, you would need to temporarily unmount the radiator and have it free floating for the test if the tubes get in the way of the GPU.
This might just be accademic as I guess you could tip the entire PC 90 degrees anticlockwise for a few seconds to flush out any air bubbles from the block into the rad whilst the PC is turned off or on (not sure, maybe both), then run another temps test.
Its not like there will be a vast amount of air as its factory filled and sealed, perhaps only 0.5, 1 or 2 ml but you wouldnt want it stuck in the block.
Bare in mind that I have never water cooled, so this is not experience but what I read and seen on vids and my understanding of basic physics.

 

[Neptune77]

At first I thought you may be on to something, but then I was looking over Steve's video again, and I saw that he had his set up exactly how I have mine for one of his "good" mounting techniques. Check out the part staring at 19:43:

At this point I'm thinking it's either a circulation issue, or that the CPU is running abnormally hot, either because of a setting, like Phaaze88 mentioned, or because this individual chip may just happen to run particularly hot.

My motherboard is an ASUS TUF Gaming Z690-Plus WiFi D4, just in case someone knows of any issues that board might be causing.

 

[jnxzi94]

I have already remounted and repasted. That improved the temps to what I am currently getting, as mentioned at the beginning of this post. As for the orientation of the pump, I originally had it mounted with the tubes coming off the right, and the logo right side up. I had a lot of trouble getting the video card in though, so I ended up changing it to what it is now.

Neither of the tubes are taking any extreme angles, so I kind of assumed that it would be fine. It isn't making any weird noises, to answer your question. Is there some way I can tell if this is causing a problem by looking at coolant temps? How much should the coolant temps normally increase during a standard 10 minute Cinebench run? As long as the coolant temps are behaving normally, shouldn't that mean that the pump tubes are fine as they are?

If nothing works, maybe it's a faulty AIO cooler. I had similar issue with Kraken M22, just to realize there is no way to fix it and CPU temp spiked super high all the time and made the CPU overheat randomly, and now I'm using Kraken X63 with no issues and with that the max. temperature under full load is 60 C. CPU is really strong i9 10850K @ 5 GHz. But I had a crackling sound which made me realize there is something badly wrong with the placement and the radiator.

 

[Neptune77]

I was considering that. It seems like the only two things that could really go wrong with an AIO other than a leak though, is the pump not working, or blocked circulation. I I know those two aspects of mine are working in some capacity, but maybe one of them isn't working as well as it should. Also, it isn't overheating randomly or anything, but it is running consistently hotter than I would expect. Do you know what exactly went wrong with your original AIO?

 

[jnxzi94]

I was considering that. It seems like the only two things that could really go wrong with an AIO other than a leak though, is the pump not working, or blocked circulation. I I know those two aspects of mine are working in some capacity, but maybe one of them isn't working as well as it should. Also, it isn't overheating randomly or anything, but it is running consistently hotter than I would expect. Do you know what exactly went wrong with your original AIO?

I was facing issues with the Kraken M22 (120mm) and I think it was due to malfunctioning (it slowly started to make this weird crackling noise after a few months, and it started to overheat the CPU, and the temperature was spiking up randomly, so it was pretty obvious that the cooling was blocked in some way).

I believe the first reason that caused this issue was that it had to be installed to the rear panel, which obviously lead to the placement being horrible.

I have a NZXT H510 Elite as my case, so now I have solved my mistake (I really shouldn't have gotten such a small cooler anyways, but I also didn't know that the positioning of these water coolers is really important) and now that I have installed the Kraken X63 (280mm) to the front panel so the placement is correct and more better and now the flow in the tubes shouldn't get blocked by anything inside.

All is good now and I'm really happy with the new AIO cooler.

But back to your problem, is your BIOS at default settings?
Is there any random high spikes in your CPU voltage?

If the cooler is malfunctioning, there is really no way to fix it, it will just die slowly and make your CPU overheat more over time.

 

[Neptune77]

I'm glad things started working for you. For my BIOS settins I'm using an XMP profile to run my ram at 3600 MHz, and I tried switching back and forth with a CPU boost setting on the main page of the BIOS, but strangely enough that didn't really seem to have an effect on the temperature. Other than that I haven't changed anything. As for the voltages, I use HWMonitor and I see that there is a voltage section for the CPU, but I don't really know anything about how to make sense of it, or what would be an abnormal value. I'll have to start reading up on that at some point.

 

[jnxzi94]

I'm glad things started working for you. For my BIOS settins I'm using an XMP profile to run my ram at 3600 MHz, and I tried switching back and forth with a CPU boost setting on the main page of the BIOS, but strangely enough that didn't really seem to have an effect on the temperature. Other than that I haven't changed anything. As for the voltages, I use HWMonitor and I see that there is a voltage section for the CPU, but I don't really know anything about how to make sense of it, or what would be an abnormal value. I'll have to start reading up on that at some point.

Can you download CPU-Z and post a screenshot of the CPU tab?
Well, it still annoys me because it was all just because of the cooling failure and I had to replace a 8 month old cooler...
RAM speed barely affects CPU temperature or anything else really.
And if your cooler doesn't make any alerting noise, it's probably fine.

 

[Neptune77]

Ok, I downloaded it. Did you want a screenshot while Cinebench is running, or just one of the CPU at idle?

 

[jnxzi94]

Ok, I downloaded it. Did you want a screenshot while Cinebench is running, or just one of the CPU at idle?

You can do both when PC is idle and when it's under 100 % load.
Also, if you can screenshot your CPU temperature too for both of the cases so I can see how everything changes.
Let's see once you post them.

 

[Neptune77]

Here are the pics...

https://imgur.com/a/zF4m9TQ

Let me know what you think.

 

[Swolley]

Here are the pics...

https://imgur.com/a/zF4m9TQ

Let me know what you think.

Erm.... Maybe it's because you're drawing 210-215w at load? Everything here seems very typical. Not sure what you expect

 

[DogNamedNibbles]

Here are the pics...

https://imgur.com/a/zF4m9TQ

Let me know what you think.

I might be a bit late on this, but it's most likely your VCore, 12700 VCore runs too high for stock and that 1.305V is too high for stock, even the worse binned 12700k's can run 4.9/3.8 - 5.0/3.9 at <1.2V. With a 5.0/4.0 - 5.1/4.0 at 1.3V - 1.25V being what you should expect.

Though, with this type of CPU, you'd be lucky to pass 60C while gaming, so even with those Cinebench temps, you should be fine.