Underclocked my 8700k to 3.7GHz lower my temps from 99°C --- Is 40°C lowest it can do?

[soulfetsh]

Cooler: H115i RGB pro XT
Ram: 32GB 3200Mhz CL16
Motherboard: ASUS Maximus Hero X Wifi
GPU:6800 XT
CPU is not delidded.

 

 

I wanted to test this after my CPU started to spike to 99°C at 5.0GHz 1.37v. When the system was stable I was getting around 78°C under load in game. Recently it started to peak at 99°C and trigger all the fans for 10 seconds every two minutes when in game. Temps are never below 85°C in game. CPU load is 80% in game. I tested 4.7Ghz at 1.35v and temps were still at around 90°C.

 

 

I downcloked the CPU to 3.7GHz at 1.15v in hopes to finally see temps below 40°C. You can see the results below 40°C under full load and idle;

https://imgur.com/a/z7L3Wui

 

 

Cooler pump is working, I can feel and hear it when it kicks in. It was replaced by Corsair 3 months ago and I was getting below 80°C under load for 2 months at 5Ghz. CPU fans are directly connected to Motherboard; not the cooler. I run them at 100% for this test to make sure I am getting best cooling. You can see the Cooler pump is at max speed. I tried resitting the cooler with different pastes. Room temp is 25°C.

 

 

Are these numbers reflective of what I should see at 3.7Ghz base speed? What could be the rise of sudden 99°C spikes at 5.0GHz ?

[LIGISTX]

7 minutes ago, soulfetsh said:

Cooler: H115i RGB pro XT
Ram: 32GB 3200Mhz CL16
Motherboard: ASUS Maximus Hero X Wifi
GPU:6800 XT
CPU is not delidded.

 

 

I wanted to test this after my CPU started to spike to 99°C at 5.0GHz 1.37v. When the system was stable I was getting around 78°C under load in game. Recently it started to peak at 99°C and trigger all the fans for 10 seconds every two minutes when in game. Temps are never below 85°C in game. CPU load is 80% in game. I tested 4.7Ghz at 1.35v and temps were still at around 90°C.

 

 

I downcloked the CPU to 3.7GHz at 1.15v in hopes to finally see temps below 40°C. You can see the results below 40°C under full load and idle;

https://imgur.com/a/z7L3Wui

 

 

Cooler pump is working, I can feel and hear it when it kicks in. It was replaced by Corsair 3 months ago and I was getting below 80°C under load for 2 months at 5Ghz. CPU fans are directly connected to Motherboard; not the cooler. I run them at 100% for this test to make sure I am getting best cooling. You can see the Cooler pump is at max speed. I tried resitting the cooler with different pastes. Room temp is 25°C.

 

 

Are these numbers reflective of what I should see at 3.7Ghz base speed? What could be the rise of sudden 99°C spikes at 5.0GHz ?

Is the water blocked mounted snugly to the mobo? Sounds like potentially a contact issue. Or your pumping too much volts through it. Is it set to 1.37v in the BIOS, is it reading that as the vcore voltage in monitor programs? What is VRVOUT being reported as?

 

Ny 8700k did 5ghz at 1.30 volts and ran relatively cool - I had a pretty good chip. 1.37 is pushing the boundary in my opinion.

 

What temps does it hit at stock speed under stress? And what programs are you using then validate stability? It’s possible your chip just wasn’t a silicon lottery winner and just simply requires too much voltage to be stable at 5ghz, and the result of pumping that voltage through it makes it way to hot.

 

My 9900k was like that. It did 5ghz at what was actually decent voltage, but it got insanely hot even at the “modest” 1.35 voltages I had it at. For a full custom loop…. And “only” 1.35v 5ghz, it would hit 99c in AIDA64 FPU stress test and thermal throttle, and would run in the low 90 in Asus real bench (which is what I would recommend you validate stability with), it was just not a great chip… it happens. 

[Eighjan]

High temp's are usually cooler issues... Like @LIGISTX said, try reseating the cooler to the CPU.

 

[OfficialTechSpace]

Massively unnecessary. Underclocking a CPU with a custom liquid cooling solution is pointless. I agree with the above, likely a contact issue.

[Mister Woof]

Contact issue or your intel toothpaste is trash and failed

 

i delidded mine 3 years ago and applied LM and it still is perfectly fine. 5ghz @ 1.38v daily usage.

[soulfetsh]

10 hours ago, LIGISTX said:

Is the water blocked mounted snugly to the mobo? Sounds like potentially a contact issue. Or your pumping too much volts through it. Is it set to 1.37v in the BIOS, is it reading that as the vcore voltage in monitor programs? What is VRVOUT being reported as?

 

Ny 8700k did 5ghz at 1.30 volts and ran relatively cool - I had a pretty good chip. 1.37 is pushing the boundary in my opinion.

 

What temps does it hit at stock speed under stress? And what programs are you using then validate stability? It’s possible your chip just wasn’t a silicon lottery winner and just simply requires too much voltage to be stable at 5ghz, and the result of pumping that voltage through it makes it way to hot.

 

My 9900k was like that. It did 5ghz at what was actually decent voltage, but it got insanely hot even at the “modest” 1.35 voltages I had it at. For a full custom loop…. And “only” 1.35v 5ghz, it would hit 99c in AIDA64 FPU stress test and thermal throttle, and would run in the low 90 in Asus real bench (which is what I would recommend you validate stability with), it was just not a great chip… it happens. 

Today I started testing with PRIME95. I replaced the thermal paste before starting today. Water block mounted snugly. Tests are done at 5GHz at 1.32v. I lowered the voltage after your reply. Idles at 38°C. Vcore reading is 1.312V in HWINFO64.

 

I will test at stock once I am done with the current tests. I didn't validate this tests yet because temps are too high.

I tested selecting the "small FFTs" option. I stopped this test when I see the results at 99C:

https://imgur.com/a/BzBl53u#8flVwL8

This is the result of another "custom" test with AVX disabled:
https://imgur.com/a/SeoJ8Si#rWbmdKF

[LIGISTX]

26 minutes ago, soulfetsh said:

Today I started testing with PRIME95. I replaced the thermal paste before starting today. Water block mounted snugly. Tests are done at 5GHz at 1.32v. I lowered the voltage after your reply. Idles at 38°C. Vcore reading is 1.312V in HWINFO64.

 

I will test at stock once I am done with the current tests. I didn't validate this tests yet because temps are too high.

I tested selecting the "small FFTs" option. I stopped this test when I see the results at 99C:

https://imgur.com/a/BzBl53u#8flVwL8

This is the result of another "custom" test with AVX disabled:
https://imgur.com/a/SeoJ8Si#rWbmdKF

Prime95 is basically a power virus, it’s not the best for testing stability anymore. 
 

Use Asus real bench: https://rog.asus.com/rog-pro/realbench-v2-leaderboard/

 

Use the stress test option, and select the amount of RAM that is the option under how much ram you have (if you have 16GB, pick 8 ). And if it passes that for the full 8 hours, your good. If not, it’s not stable. And it shouldn’t hit 99c, hopefully…. If it does, you need to dial your OC back. Also load line settings are pretty important. Too high a number will push volts higher then needed which leads to increased temps. 

[soulfetsh]

On 5/28/2021 at 12:33 PM, Mister Woof said:

Contact issue or your intel toothpaste is trash and failed

 

i delidded mine 3 years ago and applied LM and it still is perfectly fine. 5ghz @ 1.38v daily usage.

What temps are you getting? Especially under 80% or higher loads? 

[Mister Woof]

7 minutes ago, soulfetsh said:

What temps are you getting? Especially under 80% or higher loads? 

used to get at most 80c at 1.38v load voltage in something like cinebench23 loop after delid.

 

before delid, thermal throttle