[SOLVED] one core higher by 5c than the others after cpu cooler change.

Solution
You probably didn't get a completely even application of paste. There may be a spot closest to that core that is hotter where you got too much or too little coverage, or an air bubble. You can try to reapply the paste, but honestly, so long as as the hottest core is below 80°C under a FULL load running Prime95 version 26.6 (ONLY, no other version) then it probably isn't worth bothering with to do it over again. Variance between cores is normal and so long as it is below ten degrees variance from hottest to coolest core temp, you are within allowable tolerance.

If you KNOW for sure it was less than what you are seeing now, before, then trying again might be worth doing.

Arctic silver 5 is ok, fine actually, but both Noctua NT-H2 and...
You probably didn't get a completely even application of paste. There may be a spot closest to that core that is hotter where you got too much or too little coverage, or an air bubble. You can try to reapply the paste, but honestly, so long as as the hottest core is below 80°C under a FULL load running Prime95 version 26.6 (ONLY, no other version) then it probably isn't worth bothering with to do it over again. Variance between cores is normal and so long as it is below ten degrees variance from hottest to coolest core temp, you are within allowable tolerance.

If you KNOW for sure it was less than what you are seeing now, before, then trying again might be worth doing.

Arctic silver 5 is ok, fine actually, but both Noctua NT-H2 and Thermal grizzly Kryonaut are considerably better. By considerably, I mean about 3-5°C peak.

Looking at your screenshots, I only see a maximum of about two degrees difference between highest to lowest core. Not sure where you are seeing a five degree variance. Keep in mind, idle is irrelevant. Run Prime95 version 26.6, and take those same screenshots again of the core temps after about five minutes. Do not run newer versions than 26.6, that use AVX instructions. They are not steady state and they create unrealistic thermal conditions.
 
  • Like
Reactions: gikal
Solution

gikal

Prominent
Apr 24, 2019
38
2
535
You probably didn't get a completely even application of paste. There may be a spot closest to that core that is hotter where you got too much or too little coverage, or an air bubble. You can try to reapply the paste, but honestly, so long as as the hottest core is below 80°C under a FULL load running Prime95 version 26.6 (ONLY, no other version) then it probably isn't worth bothering with to do it over again. Variance between cores is normal and so long as it is below ten degrees variance from hottest to coolest core temp, you are within allowable tolerance.

If you KNOW for sure it was less than what you are seeing now, before, then trying again might be worth doing.

Arctic silver 5 is ok, fine actually, but both Noctua NT-H2 and Thermal grizzly Kryonaut are considerably better. By considerably, I mean about 3-5°C peak.

Looking at your screenshots, I only see a maximum of about two degrees difference between highest to lowest core. Not sure where you are seeing a five degree variance. Keep in mind, idle is irrelevant. Run Prime95 version 26.6, and take those same screenshots again of the core temps after about five minutes. Do not run newer versions than 26.6, that use AVX instructions. They are not steady state and they create unrealistic thermal conditions.
thank you sir, i will do that tomorrow and i will show you the results :D, i wanted to get kryonaut but is extremely rare here in colombia so im going to order one for the next thermal paste replacement
 
as one core typically runs at turbo speeds, and the other cores at lower speed , I'd expect to see one core warmer...

You'd need to verify with HWMonitor when all cores are at equal speed and load (Prime 95/small FFT) that one core is truly warmer or not...
 
  • Like
Reactions: gikal

gikal

Prominent
Apr 24, 2019
38
2
535
as one core typically runs at turbo speeds, and the other cores at lower speed , I'd expect to see one core warmer...

You'd need to verify with HWMonitor when all cores are at equal speed and load (Prime 95/small FFT) that one core is truly warmer or not...
where can i get this program free of adware or malware D:? thanks
 
Actually, one core doesn't "typically" run at turbo speeds. In fact, it's fairly rare and in practice, extremely difficult, to get just one core to run under boosted load conditions. It happens, but rarely, and only if you are using a completely linear optimized piece of software which is getting harder and harder to find as practically all modern applications and game engine are at least somewhat optimized for multithreaded processing. Even those that are usually won't cause that to happen because there are too many other processes requiring use of CPU resources for it to be likely.

The operating system and all it's continuously running processes need access to core resources in addition to whatever game, benchmark or application you are running, so about the only way you're going to get a single core to run by itself, under any kind of demanding load, is to disable all the other cores. It simply isn't likely to be seen under normal circumstances. If the system is under a demanding load, of any kind, it is far more likely that you will at least see 2-4 cores boosting, and usually, all of them will take turns OR go full all core boost.
 
Take some pics and post them here. I have no idea without seeing. Likely, it is old thermal paste, but I can't say that without seeing it for sure.

You'll need to host the images at a site like imgur.com or tinypic.com, then copy the link for "direct layouts" and then come here, click on the picture button above your post on the formatting toolbar and then paste the code you copied from the hosting site.
 
  • Like
Reactions: gikal

gikal

Prominent
Apr 24, 2019
38
2
535
Take some pics and post them here. I have no idea without seeing. Likely, it is old thermal paste, but I can't say that without seeing it for sure.

You'll need to host the images at a site like imgur.com or tinypic.com, then copy the link for "direct layouts" and then come here, click on the picture button above your post on the formatting toolbar and then paste the code you copied from the hosting site.
hello sir, this are my tests temperatures after 5 min (room temperature is 16c :
hwmonitor: View: https://imgur.com/a/Yu9yjQX

hwinfo64: View: https://imgur.com/a/UpHRlXf


and this is the img you asked, thank you so much for helping me :D
View: https://imgur.com/a/tlfF6m3
 
Those temps are fine. There is no problem with a 4 degree differential. I wouldn't bother to repaste unless you really feel the need to. As far as your pic of the CPU cooler standoff, you're going to need a MUCH better image than that. I can't tell anything from that image.
 
  • Like
Reactions: gikal

TRENDING THREADS