Core #1 always hotter

[AverageLeo]

I have a weird problem,

my #1 (not #0) is always hotter than the other cores.

I heard a lot of applications run as single-threaded but in this case I'm running full-threaded apps like Cinebench and other stress-tests.

I saw in some forums that the first core is leaning to be hotter but in my case it's the second core (#1, not #0).

BTW I changed the thermal paste today and re-installed the AIO cooler.

 

Thoughts? Tips?

Thanks

 

https://gyazo.com/e273b24d44e81f2fe4705069ff07d2b1

[5x5]

It's probably cause of the paste - variance is normal for Intel's paste-based CPUs

[AverageLeo]

3 minutes ago, 5x5 said:

It's probably cause of the paste - variance is normal for Intel's paste-based CPUs

Paste was installed pretty good imo and it's a good paste from Noctua as well.

[Radium_Angel]

21 minutes ago, AverageLeo said:

hotter

Unless you are hitting thermal limits, I wouldn't worry about it

[thedangerine]

Could just be the location it is on the die, my cousins 3rd get i5 had one core that was always a little hotter because it was in between two other cores and got extra heat from them

[schwellmo92]

3 minutes ago, AverageLeo said:

Paste was installed pretty good imo and it's a good paste from Noctua as well.

He's talking about the paste under the IHS.

[AverageLeo]

2 minutes ago, Radium_Angel said:

Unless you are hitting thermal limits, I wouldn't worry about it

I want to OC but I can't reach the top if one of the cores is hot.

[Radium_Angel]

Just now, AverageLeo said:

I want to OC but I can't reach the top if one of the cores is hot.

You could always delid if you are skilled enough

[AverageLeo]

On 4/24/2020 at 5:34 PM, schwellmo92 said:

He's talking about the paste under the IHS.

So delidding will fix it?

 

On 4/24/2020 at 5:37 PM, Radium_Angel said:

You could always delid if you are skilled enough

I don't have metal liquid.. Good paste is enough?

[Radium_Angel]

Just now, AverageLeo said:

I don't have metal liquid.. Good paste is enough?

Beats me, I don't delid my Xeons, it was just an option if you are unsatisfied with the heat of your CPU

[5x5]

22 minutes ago, AverageLeo said:

Paste was installed pretty good imo and it's a good paste from Noctua as well.

I'm talking about the Intel paste under the IHS

 

15 minutes ago, AverageLeo said:

So delidding will fix it?

no, it won't. unless you can use solder, no paste will fix this. Just ignore it

[schwellmo92]

24 minutes ago, AverageLeo said:

So delidding will fix it?

If you use liquid metal yes, but the temps are fine I wouldn't worry about it who cares if it hits 72c or something.

[AverageLeo]

1 hour ago, schwellmo92 said:

If you use liquid metal yes, but the temps are fine I wouldn't worry about it who cares if it hits 72c or something.

I'm trying to OC to 90c and I can't if one core reaches 100c.

[schwellmo92]

28 minutes ago, AverageLeo said:

I'm trying to OC to 90c and I can't if one core reaches 100c.

Try the OC first, if it then becomes a problem the only solution is a delid with liquid metal and removing the silicon around where the IHS joins the CPU PCB.

[Falkentyne]

Even numbered cores being hotter is standard fare.  Happens on 8700k and 9900k also.  Cores are convex that's why.

Thicker thermal paste, LM, or heatsink mods or IHS mods can help a bit.  Sanding the slug down doesn't fix the problem, oddly enough, but it does bring the deltas closer together.

[Mister Woof]

Even with a delid w/LM, there's some core to core variance.

 

After 15 minutes on RealBench:

 

0: 82

1: 77

2: 82

3: 78

4: 81

5: 79

 

Granted my cooler is only modest at best. There's still a 5c variance from coolest to hottest.

[AverageLeo]

20 minutes ago, Mister Woof said:

Even with a delid w/LM, there's some core to core variance.

 

After 15 minutes on RealBench:

 

0: 82

1: 77

2: 82

3: 78

4: 81

5: 79

 

Granted my cooler is only modest at best. There's still a 5c variance from coolest to hottest.

I have 10-12c diffrenece on the high temps while benching/stressing

[Mister Woof]

9 minutes ago, AverageLeo said:

I have 10-12c diffrenece on the high temps while benching/stressing

And I've delidded my CPU and applied liquid metal - just trying to give some perspective on how even with better circumstances, there is still variance.

[AverageLeo]

1 hour ago, Mister Woof said:

And I've delidded my CPU and applied liquid metal - just trying to give some perspective on how even with better circumstances, there is still variance.

you talk about 5c I''m talking about 15c.

[Mister Woof]

2 minutes ago, AverageLeo said:

you talk about 5c I''m talking about 15c.

 

The factory paste seems to be applied by a glass sorter, and delidding and applying LM fixes that (mostly, in my case, 5c variance).

 

In your case, the CPU is 3 years old and the poorly applied paste can be dried up. Decapitate that fucker and go to down, IMO.

[AverageLeo]

11 minutes ago, Mister Woof said:

 

The factory paste seems to be applied by a glass sorter, and delidding and applying LM fixes that (mostly, in my case, 5c variance).

 

In your case, the CPU is 3 years old and the poorly applied paste can be dried up. Decapitate that fucker and go to down, IMO.

Delid tool is 50$ and the liquid is another 15$.. I'm not sure it's worth the try.

What do you think?

[Mister Woof]

4 minutes ago, AverageLeo said:

Delid tool is 50$ and the liquid is another 15$.. I'm not sure it's worth the try.

What do you think?

i mean, if those are just under synthetic stress tests and you're okay with the frequency, you're fine if you're under 100. You won't see those temps in real world general use.

[Guest]

Delid it. 

Dont need to buy the tool. You can use the vice out in the garage or just use a razor and cut the glue.

 

You'll have the chip back in the PC and running within an hour. 

 

If you are going to be serious and delid, Woof can tell you everything you need to know. 

 

For a tech difficulty rating 1-10, maybe a 4. After you've done a few, level 1 easy as eating pie.

[AverageLeo]

45 minutes ago, Mister Woof said:

i mean, if those are just under synthetic stress tests and you're okay with the frequency, you're fine if you're under 100. You won't see those temps in real world general use.

I'm trying to reach a stable 5.0ghz and I can't becasue of this damn core.

My settings:

Ai Tuner - XMP
XMP - DDR4 3000
BCLK - 100.00
AVX neg offset - 0
CPU ratio limit - 50
CPU cache ratio min - Auto
CPU cache ratio max - 45 / 44 / Auto42
CPU Load Line Cal - Level 6 (tried Level 5 as well)
VRM Spread Spectrum - Disabled
IA AC load line - 0.01 (tried Auto as well)
IA DC load line - 0.01 (tried Auto as well)
BCLK Adaptive Voltage - Enabled
CPU Core/Cache Voltage - Manual
Core Voltage Override - 1.325overheat / 1.320crash / 1.310crash / 1.300crash
DRAM Voltage - 1.353
CPU VCCIO Voltage - 1.100 (tried Auto as well)
CPU System Agent Voltage - 1.150 (tried Auto as well)

 

It came to a situation that all my cores are around 85c and core #1 is 100c.

Should I delid with a razor? and I have to use liquid metel right?

[Mister Woof]

4 minutes ago, AverageLeo said:

I'm trying to reach a stable 5.0ghz and I can't becasue of this damn core.

My settings:

Ai Tuner - XMP
XMP - DDR4 3000
BCLK - 100.00
AVX neg offset - 0
CPU ratio limit - 50
CPU cache ratio min - Auto
CPU cache ratio max - 45 / 44 / Auto42
CPU Load Line Cal - Level 6 (tried Level 5 as well)
VRM Spread Spectrum - Disabled
IA AC load line - 0.01 (tried Auto as well)
IA DC load line - 0.01 (tried Auto as well)
BCLK Adaptive Voltage - Enabled
CPU Core/Cache Voltage - Manual
Core Voltage Override - 1.325overheat / 1.320crash / 1.310crash / 1.300crash
DRAM Voltage - 1.353
CPU VCCIO Voltage - 1.100 (tried Auto as well)
CPU System Agent Voltage - 1.150 (tried Auto as well)

 

It came to a situation that all my cores are around 85c and core #1 is 100c.

Should I delid with a razor? and I have to use liquid metel right?

Delid is probably your only option to go past 1.325v. There's no guarantee you can get 5ghz stable even if you delid, though. 5ghz on a 7700k isn't very common. You may want to settle on 4.7-4.8ghz at a more reasonable voltage. This is with or without delid. Even if you don't overheat at 1.325 after the delid, it could still crash after a test.

 

I haven't tried delidding with a razor, as I used a tool. It doesn't seem that difficult, though, and there's lots of videos on youtube. Liquid metal will get you the best results, but you can use whatever. The intel stock paste is poorly applied and of poor quality to begin with, so pretty much anything is an upgrade.