Questions arising from overclocking 9600K related to the H100i PRO AIO

[andiey]

I have found another problem. Today's weather dropped. The room temperature has dropped by 3 degrees. But when I leave the system @ stock speed idle, the temperature is higher than all the stress tests before.
I imagine that the thermal paste comes with the AIO has melted and unevenly spreaded. Another evidence is that 2 out of 6 cores always differ from the other by 8 to 10 degrees, is that a sign that the thermal needs to be replaced?

 

[EarthDog]

Are you saying that now at idle your cpu temperature is higher than on load??? Whaa? That makes no sense.

Core temperatures are typically of by several degrees. That is normal. Your thermal paste is likely fine. Though if you have extra paste, you can pull it and see, wipe off and apply the new.

 

[andiey]

No no
I am saying that when I first started running the machine, I don't do any stress test. The temperature at idle were like 35, 36 somewhere in the 30's no 40's.

Recently, I have ran several times stress test, also I have overclocked the CPU and ran the stress test.

Now I undo everything and back to stock speed.
Today under 22 room temperature, I am getting some 40s from one of the cores at ilde.
Also, out fo the 6 cores, there is always one of them hit mid to high 40s.
I am just worried if the thermal paste (comes with the cooler) is somewhat less functional. I have this suspicions is because of my old machine. I cleaned it. Applied new thermal paste to it (Thermaltake TG2) and all 4 cores of the 3570K are around 31 at ilde. All cores temp are even

 

[Zach_01]

Is the AIO’s water temp decreased also along with ambient?

If yes, when core temps increase, could indicate low conductivity between CPU IHS and water block...

If water temp increase with core temps, then maybe you need to readjust the fan curve.

 

[Deleted member 185158]

No no
I am saying that when I first started running the machine, I don't do any stress test. The temperature at idle were like 35, 36 somewhere in the 30's no 40's.

Recently, I have ran several times stress test, also I have overclocked the CPU and ran the stress test.

Now I undo everything and back to stock speed.
Today under 22 room temperature, I am getting some 40s from one of the cores at ilde.
Also, out fo the 6 cores, there is always one of them hit mid to high 40s.
I am just worried if the thermal paste (comes with the cooler) is somewhat less functional. I have this suspicions is because of my old machine. I cleaned it. Applied new thermal paste to it (Thermaltake TG2) and all 4 cores of the 3570K are around 31 at ilde. All cores temp are even

Can't compare the processors for thermal dynamics. So you need to put old rig behind you and start understanding your new rig.
People here are pointing you in the right direction. Core difference activity is pretty normal. No worries.
Max temp from your first post 64c and that's a very good temperature. I don't feel you need to re-apply the thermal interface material.

When your temps get to 95c, then you're getting hot. you have 30c headroom.



additional comment:
Only your load temp really matters. Don't be terribly concerned with idle temps unless they are crazy high like 60-70c range.

 

[jaggerwild]

I hope when overclocking your not using the auto Asus Software, also when running stress tests(small FFT or VVT or whateva)I have seen it keep running even after I have shut it off(FYI).

 

[andiey]

Folks,

I have a problem!

I remove the thermal paste and reapplied my own which is Thermaltake TG-2.
Now the temperature is slightly higher than before.

All at stock speed, I left the machine COMPLETELY IDLE only running HWiNFO for 2 hours.

Core #1 MAX 54C
Core #2 MAX 54C
Core #3 MAX 54C
Core #4 MAX 52C
Core #5 MAX 53C
Core #6 MAX 55C

I might have applied a little bit thicker than I did to my 3570K a few days ago. The 3570K's thrmal reading is very even.
I also suspect would that be the hard plastic CPU bracket that hinders the pump to have a good contact with the CPU.

Should I try re-pasting thermal one more time to see if there are any luck?

 

[oxrufiioxo]

Folks,

I have a problem!

I remove the thermal paste and reapplied my own which is Thermaltake TG-2.
Now the temperature is slightly higher than before.

All at stock speed, I left the machine COMPLETELY IDLE only running HWiNFO for 2 hours.

Core #1 MAX 54C
Core #2 MAX 54C
Core #3 MAX 54C
Core #4 MAX 52C
Core #5 MAX 53C
Core #6 MAX 55C

I might have applied a little bit thicker than I did to my 3570K a few days ago. The 3570K's thrmal reading is very even.
I also suspect would that be the hard plastic CPU bracket that hinders the pump to have a good contact with the CPU.

Should I try re-pasting thermal one more time to see if there are any luck?

if you had ihs contact issues during a prime or aida extreme cpu/fpu stress test you'd likely hit TJ max... idle peak temps are pretty irrelevant. what is your 100% all core load temps. Try running cinebench R20 and set the minimum test duration to 1200 seconds.

 

[andiey]

what is your 100% all core load temps

You want the max temp or the average?

 

[oxrufiioxo]

You want the max temp or the average?

max during a real stress test not cpuz though it doesn't stress a cpu hard enough.

you should also run your cooler on at least the balanced and preferably the extreme profile when stress testing.

 

[andiey]

some boards have tweaked things when using xmp... like asus. When you enable xmp, it tells you specifically about core enhancement. You chose to enable it. You can disable it and overclock manually. This way you can control voltage and perhaps lower the heat as leaving core voltage on auto tends to use more than is needed.

I also notice that about th auto voltage. When overclocking I remember the magic number 1.35V is being the most for the CPU and I saw it jumped to 1.37V intermittently.
The trouble is I don't know how to manually overclock it.

 

[EarthDog]

I also notice that about th auto voltage. When overclocking I remember the magic number 1.35V is being the most for the CPU and I saw it jumped to 1.37V intermittently.
The trouble is I don't know how to manually overclock it.

Voltage left on auto

Learn.There are plenty of guides out there if you take the time to look. Likely some here at the site.

 

[andiey]

max during a real stress test not cpuz though it doesn't stress a cpu hard enough.

Can I use CPU-Z's bench? I know Prime95 and Aiada 64 are too much for a new CPU

 

[EarthDog]

I know Prime95 and Aiada 64 are too much for a new CPU

New or old is absolutely not relevant. THey are more stressful as they use AVX instructions. I find AIDA64, the default stress test, does fine for me for my uses.

 

[oxrufiioxo]

Can I use CPU-Z's bench? I know Prime95 and Aiada 64 are too much for a new CPU

according to who.... I use aida 64 extreme fpu to test stability on every new cpu I have purchased or built for someone else for almost a decade

New or old is absolutely not relevant. THey are more stressful as they use AVX instructions. I find AIDA64, the default stress test, does fine for me for my uses.

I agree, even if you disable FPU it's still better than CPUZ

 

[EarthDog]

I run it at default but when I really want to kick the CPU in the pants, I use FPU only (temps similar to P94 Small FFT).

 

[oxrufiioxo]

@andiey if you're not comfortable manual OC'ing you're better off at best running MCE as any form of auto oc other than that is going to run your voltages way too high.

Also even with a 2080 ti assuming you're using your system for gaming the gains aren't worth it imo




if you're just trying to learn then I would suggest watching this video even though the cpu is different a lot of the settings still apply to your motherboard.

 

[andiey]

Things are developing and not looking good.

I THINK I HAVE SPOILED MY NEW CPU!!!

I re-pasted the thermal compound yesterday and ran 2 times with stock and overclock.

Just now I ran it once at stock, and here are the results
55 55 56 55 54 55 (MAX Core 1, 2, ...5, 6)
48, 48, 49, 48, 47, 48 (AVG Core 1, 2,...,5, 6)

The average is higher than yesterday, I guess the thermal paste has settled down after over 24 hours. Today is very cold outside, my room temp also decreases to 19.5 degree. Yesterday was 24 degree.

OMG, I haven't ran any games yet, now I am worrying if they will shoot up to over 70C even talking about stock speed.
Look, I am not intending to run overclock as daily, I m just curious about the 9600K' potential and want to know if this die can go up to 5GHz. To be honest, I will go for 9700K when the price drops hopefully next year,


Right now, I just want to get back what I got before, a cooler CPU. I was getting a max of 48 only before all this overclock testings and re-paste. I shouldn't have overclocked!

This is my old 3570K thermal, the max is only 38 Celsius @ 4.2GHz (Overclocked with XMP#1)
The 9600K 56 Celsius is much higher and it's only 100MHz higher than the 3570K??!!

 

[EarthDog]

I dont see a problem with the temps...just a knowledge gap.

You cannot compare these CPUs. There are so many differences which affect temps it just isn't possible. Again, I don't see an issue here outside of misinformed assumptions driving these worries.

The temps you are seeing, this is from a stress test application? If so, gaming is typically LESS than any stress test (see above about assumptions based on bologna).

 

[andiey]

The temps you are seeing, this is from a stress test application? If so, gaming is typically LESS than any stress test (see above about assumptions based on bologna).

The 9600K values are obtained by just leaving the system at idle state.

The attached screenshot of the overclocked 3570K shows temp when I was doing just browsing internet.

 

[EarthDog]

The 9600K values are obtained by just leaving the system at idle state.

The attached screenshot of the overclocked 3570K shows temp when I was doing just browsing internet.

Idle temps, as was told to you previously, are not terribly relevant. Please, let's move on from the past and a chip you cannot compare your new one to and see what actual load temps are in a stress test. I suggest running Aida64 at its default settings and then post up a screenshot of the results. In general we state to keep these CPUs under 90C when stress testing. They will throttle and protect themselves at 100C IIRC.

To put it simply your new CPU does NOT run cooler than your old one. It is what it is.

 

[oxrufiioxo]

You should also post a screenshot of your 9600k system.



Also you seem overly focused on thermal paste it's not what's causing you issues unless you didn't use any at all

 

[andiey]

This is the result of the 3570K. The peak temp of all 4 cores during the test are 63, 66, 65, 65.

I can run it on 9600K but given it is so stressing, will my 9600K be damaged?

Idle temps, as was told to you previously, are not terribly relevant. Please, let's move on from the past and a chip you cannot compare your new one to and see what actual load temps are in a stress test. I suggest running Aida64 at its default settings and then post up a screenshot of the results. In general we state to keep these CPUs under 90C when stress testing. They will throttle and protect themselves at 100C IIRC.

To put it simply your new CPU does NOT run cooler than your old one. It is what it is.

OK

 

[oxrufiioxo]

We don't care about your 3570k it has nothing to do with your 9600k. One is an old 22nm CPU that runs extremely cool at stock/xmp settings and the other is a 14nm CPU that has a much higher thermal density.


Your system will shut down before anything bad happens unless you manually went in and set voltage way too high.

 

[EarthDog]

This is the result of the 3570K. The peak temp of all 4 cores during the test are 63, 66, 65, 65.

I can run it on 9600K but given it is so stressing, will my 9600K be damaged?


OK

1. Dont care about your 3570k.
2. No it wont hurt it. Did you see what I said above about temps and throttling?
3. You ran a GPU test... not the cpu test... come on man!
Edit: run aida64 stress test. Nit the gpgpu test (which does have a cpu component)

Obviously we want to see TEMPERATURES not how fast it ran it. Stick with us....