Weird temps on i7 4790k

[roylapoutre]

Hi guys,

 

I'm dealing with something I've never saw before, either on my current CPU or my previous ones. I re-applied thermal paste (Artic MX-2, the same one I had before) on my CPU, like I always do with a cross pattern to be sure it spreads everywhere, and when stress testing the CPU, I noticed extremely high load temperatures (~90-100°C) on the first core while getting 60-70°C on the other ones. This was not the case before. I tried OCCT, Prime95, Cinebench and a bit of video rendering, same pattern everytime. In gaming (30-40% load scenarii), the first one can get a little hotter than the others, but it stays below 75°C. At idle, all cores are always around 30°C and the first one usually 3 or 4° higher which is perfectly normal. I looked at temps with CAM, OCCT, HWMonitor and SpeedFan.

 

I obviously thought of an unevenly re-mounted cooler as well as unevenly applied thermal paste so I tried to do all of this carefully, 3 times. No changes. Great idle temps and one core going crazy on load. I know the thermal paste is applied correctly, I know the cooler is mounted correctly, so I wanted to know if someone had a magical idea to help me understand that crazy behavior. I thought of a defective detector, but I can't verify this hypothesis. Any input would be appreciated, thanks

 

EDIT : oh and I forgot, I'm running it at 4.6GHz and 1.2V

[PCGuy_5960]

It is logical... You have an AIR cooler not a Watercooler....

[PCGuy_5960]

Prime95 is also designed to stress the CPU as far as it goes...

You can try AIDA64 and RealBench to verify your results and do a Real-World test...

Run cinebench multiple times, render a video in After Effects etc.

[Nacho Marco Segui]

2 minutes ago, PCGuy_5960 said:

It is logical... You have an AIR cooler not a Watercooler....

That is not the issue... you don't need a watercooler for a 4790k at 1.2v....

[Nacho Marco Segui]

Maybe the paste under the IHS is not covering the die properly. On my 4790k,cores 3 and 4 are always 2 to 4 degrees higher.

[PCGuy_5960]

1 minute ago, Nacho Marco Segui said:

That is not the issue... you don't need a watercooler for a 4790k at 1.2v....

If it was an enthusiast grade chip (5820K,5930K,5960X,6800K etc) it shouldn't be a problem as the IHS is soldered...

On the 4790K Intel is using a thermal compound to glue the IHS to the CPU die...

The OP should also check his fan speed...

[roylapoutre]

1 minute ago, Nacho Marco Segui said:

Maybe the paste under the IHS is not covering the die properly. On my 4790k,cores 3 and 4 are always 2 to 4 degrees higher.

The thing is I've never noticed that behavior before. The first core were always on par with the others and it never got past 80°C in extreme cases. I'm really puzzled. Oh and the CPU doesn't even throttle when it's hitting 95°C which is kinda odd

[Nacho Marco Segui]

1 minute ago, PCGuy_5960 said:

If it was an enthusiast grade chip (5820K,5930K,5960X,6800K etc) it shouldn't be a problem as the IHS is soldered...

On the 4790K Intel is using a thermal compound to glue the IHS to the CPU die...

The OP should also check his fan speed...

Thanks, I guess. I already knew that.

[roylapoutre]

2 minutes ago, PCGuy_5960 said:

If it was an enthusiast grade chip (5820K,5930K,5960X,6800K etc) it shouldn't be a problem as the IHS is soldered...

On the 4790K Intel is using a thermal compound to glue the IHS to the CPU die...

The OP should also check his fan speed...

Fan speed are ok, full RPM. And it wouldn't affect only one core if the fans weren't spinning fast enough

[Nacho Marco Segui]

1 minute ago, roylapoutre said:

The thing is I've never noticed that behavior before. The first core were always on par with the others and it never got past 80°C in extreme cases. I'm really puzzled. Oh and the CPU doesn't even throttle when it's hitting 95°C which is kinda odd

Then it is probably the sensor, but it is strange that it doesn'y throttle. Now I am puzzled just like you haha

[PCGuy_5960]

1 minute ago, Nacho Marco Segui said:

Thanks, I guess. I already knew that.

What I meant is that maybe the thermal compound behind the IHS has hardened over time and is causing thermal resistance...

[roylapoutre]

3 minutes ago, Nacho Marco Segui said:

Then it is probably the sensor, but it is strange that it doesn'y throttle. Now I am puzzled just like you haha

Ahah that's both reassuring and frightening.

 

Just did it again, the first core goes as far as 97°C on Cinebench while the second hotter core is at 72°C... Really WTF

[roylapoutre]

4 minutes ago, PCGuy_5960 said:

What I meant is that maybe the thermal compound behind the IHS has hardened over time and is causing thermal resistance...

Well, I bought the CPU brand new 2 months ago soooo  

[PCGuy_5960]

1 minute ago, roylapoutre said:

Well, I bought the CPU brand new 2 months ago soooo  

It doesn't matter... If it was sitting on a shelf for a year this might be the case... Make sure that Prime95 is not overvolting your CPU! In some older versions it did and this caused thermal issues...

[Nacho Marco Segui]

6 minutes ago, PCGuy_5960 said:

What I meant is that maybe the thermal compound behind the IHS has hardened over time and is causing thermal resistance...

That is exactly what I said in my first post, the paste is not covering the whole die

[Nacho Marco Segui]

3 minutes ago, roylapoutre said:

Well, I bought the CPU brand new 2 months ago soooo  

Does not matter, maybe it is a mistake from the factory. Gotta delid haha

[roylapoutre]

2 minutes ago, PCGuy_5960 said:

It doesn't matter... If it was sitting on a shelf for a year this might be the case... Make sure that Prime95 is not overvolting your CPU! In some older versions it did and this caused thermal issues...

I checked carefully the voltage/frequency with different software, and it's constantly locked at 1.2V/4.6GHz during load !

[PCGuy_5960]

1 minute ago, roylapoutre said:

I checked carefully the voltage/frequency with different software, and it's constantly locked at 1.2V/4.6GHz during load !

Then you have 4 options:

1. Lower the Voltage and Frequency
2. Buy at least a 240mm Watercooler
3. Delid the CPU
4. Return it and ask for a replacement

[roylapoutre]

2 minutes ago, Nacho Marco Segui said:

Does not matter, maybe it is a mistake from the factory. Gotta delid haha

I found it hard to believe that factory malfunction can suddenly appear 2 months after usage but why not 

[roylapoutre]

2 minutes ago, PCGuy_5960 said:

Then you have 4 options:

1. Lower the Voltage and Frequency
2. Buy at least a 240mm Watercooler
3. Delid the CPU
4. Return it and ask for a replacement

I shouldn't need the first two options in my case, so I might need to return it, don't want to delid it while it's under warranty, I'm not THAT crazy 

[PCGuy_5960]

Give AIDA64 a shot... I think it is the best stress testing software...

http://www.aida64.com/downloads

Realbench is also good and it is a real world stress testing software

http://rog.asus.com/19932016/news/realbench-v2-43-new-version-available-now/

[roylapoutre]

2 minutes ago, PCGuy_5960 said:

Give AIDA64 a shot... I think it is the best stress testing software...

http://www.aida64.com/downloads

Realbench is also good and it is a real world stress testing software

http://rog.asus.com/19932016/news/realbench-v2-43-new-version-available-now/

Oh yeah, forgot AIDA64, it's much more realistic in its way of stressing the CPU. But at that point, I would be surprised to see much of a change

[PCGuy_5960]

1 minute ago, roylapoutre said:

Oh yeah, forgot AIDA64, it's much more realistic in its way of stressing the CPU. But at that point, I would be surprised to see much of a change

Anyway, try them and send us a screenshot... Use RealTemp or HW Monitor for Max temp...

[roylapoutre]

Here's a screenshot taken during a run of Realbench. It seems to impact the first core as much as the other stress tests.

[roylapoutre]

And idle temps for the sake of it