Intel i7 2700K OC heat problems

How bad are the temperatures on full load?

  • Upper limit

    Votes: 0 0.0%
  • You will damage your CPU like this

    Votes: 0 0.0%

  • Total voters
    0

Parleoth

New Member
Dear Community

I started overclocking my CPU only with the mainboard (Asus p8z68-v pro gen3) and ended up with crazy high temperatures. I assume that my idle temps are already too high even tho all vents work just fine, I just applied new heat transfer paste and the case is positioned right next to the window and thus the airflow is not blocked by anything.

My setup for the test is the following:

Intel Core i7-2700K (base clock 3.5 GHz)
Corsair H70 watercooling for CPU

In the following picture you see the temperatures after a cinebench benchmark testrun (stock clock):
(see Stock.png)

Then I used the already pretuned profil for OC that the mainboard provides and after the same testrun the temperatures look like this (4 GHz overclocked):
(see OC.png)

Anyone has an idea what could be the issue?
 

Attachments

  • OC.png
    OC.png
    38.7 KB · Views: 12
  • Stock.png
    Stock.png
    46.5 KB · Views: 11

beers

Moderator
Staff member
What are your OC parameters?

Stock doesn't look bad.

Make sure the pump isn't on a PWM configured header.
 

Parleoth

New Member
What are your OC parameters?

Stock doesn't look bad.

Make sure the pump isn't on a PWM configured header.

I just used a predefined OC profil that the mainboard offered. Instead of all auto, there is another profil that boosts it to 4 GHz. You could also another OC mode but then I get temperatures close to 100°C which I consider much too high.

About the PWM header. I never heard of that before? I know PWM but didn't know there are such headers on the mainboard? The H70 is on the highest level all the time
 

Jiniix

Well-Known Member
You can see my i7-2600K in my signature. It's overclocked to 4.4GHz, and with synthetic stress tests (generates an unrealistic load) the max temp would be around 75-80C on a hot day.
Are you using auto voltage? Can't see any images or info about that. IIRC built in auto-OC would do 4.3GHz at almost 1.4v, while I did 4.4GHz at 1.275v.
 

Parleoth

New Member
You can see my i7-2600K in my signature. It's overclocked to 4.4GHz, and with synthetic stress tests (generates an unrealistic load) the max temp would be around 75-80C on a hot day.
Are you using auto voltage? Can't see any images or info about that. IIRC built in auto-OC would do 4.3GHz at almost 1.4v, while I did 4.4GHz at 1.275v.

That is very interesting! What is your idle temperature of the CPU? Are all cores equally hot? I have differences up to 6°C between the cores...
Yes I use auto voltage. On stock everything is on auto and even on the OC profile most of the settings are on auto.You can see the OC profile specs in the picture below
But at 4.4 GHz I get crazy high temperatures. Any suggestions how to downclock it a little to get temperature <80°C?
 

Attachments

  • Unbenannt.jpg
    Unbenannt.jpg
    159 KB · Views: 5

Jiniix

Well-Known Member
The built-in OC profile is honestly pretty garbage. As you can see it's using a BCLK of 102, which can cause system insabilities, and in total nets you 0.09GHz extra.
For temps on a CPU, it's pretty much 99% down to the voltage, and not the frequency. A variation from core to core of 6C is pretty normal.

My friend is borrowing the machine at the moment, I've upgraded to an i7-8700K build, just haven't updated signature yet. But I would say idle temps were in the 35-45 range.
I would try to the BCLK/PCIE freq to 100, change the voltage type from "Offset" to "Manual" and type in 1.3v
Should be stable at 4.3GHz, and you can stress test to see if the temps drop.
 

Parleoth

New Member
I tried that out Base clock of 100 MHz, a multiplier of 43 and a manual voltage of 1.3V. During a stress test my temps went up to 92-98°C so the CPU started reducing the clock to 4.2 GHz. Should I reduce the voltage even more or play a little with the multiplier?
Idle temp now is about 52°C instead of the 45° stock
 

Jiniix

Well-Known Member
You could try a more conservative overclock, with 1.25v and a 42 multiplier. But those temperatures seem very high regardless.
 

Darren

Moderator
Staff member
Is your H70 making any kind of sound? The pump on my H100 quit and wouldn't cool properly under load anymore. Had a slight rattle/grinding noise for a few months before it finally quit completely. As old as your H70 is I'm surprised its still working. 5 years is usually the max on water coolers from that era.
 

Intel_man

VIP Member
I'd replace the h70, it's a mediocre cooler when it first came out and being that it is 6 years old now, the performance of that cooler deteriorated to what you're seeing now.
 

Parleoth

New Member
@Jiniix
1.25V with 4.2 GHz works pretty well. Under a stress test the temp was <90°C so I will keep it like that for a while.

@Darren
The pump sounds just fine. No noise that appears when there're air bubbles or so. I am also surprised how long my equipment lasts. Unfortunately I never measured the temperature when it was brand new but it seems okay for a 6 y/o cooler unit.

@Intel_man
If I start changing anything apart from the GPU, I'll just build a whole new computer. I try to get the most out of it to make it deal with newer games for another year or so and when the RTX 20XX serie really shows its value, I might build a new one with a 2080 (Ti).
 

Darren

Moderator
Staff member
I know Intels run hotter than AMD chips but really anything over 70oC makes me uncomfortable. My motherboard is set to shut off at 75oC. 90 just seems dangerous.
 

Jiniix

Well-Known Member
Agree with Darren, 80-85C is the temperature I use as a worst case ceiling. Stress testing on a warm summer day, while in reality in games / heavy workloads it should be around 65-75C tops.
 

Jiniix

Well-Known Member
Is that something you've set manually? 75 seems pretty low for automatic thermal shutdown.
Not something I would set for sure. Even on brand new motherboards, there's always one sensor doing something erratic.
On my Z370 board there's a temp sensor that either reads 0C or 128C, nothing in between. Obviously wrong, but could trigger a shutdown I assume.
 

Darren

Moderator
Staff member
Is that something you've set manually? 75 seems pretty low for automatic thermal shutdown.
Yup. Pretty sure it's 75 or 80. Better safe than sorry and it's never even really gotten close even at 4.0GHz under full load. My cooler has a TDP design of 250w and my chip is less than half that.

My sensors are pretty solid on this board, there's some confusion on the temps for the CPU vs CPU socket vs CPU package but that's an AMD thing since the reported CPU temp is more of a calculation than an actual reading.
 
Top