Ryzen 6 3600 CPU's Sudden Temperature Fluctuations with SPC Fortis 3 v2 Cooling

Hello

Today I put together a new setup and am puzzled by the sudden temperature spikes on the CPU. Ryzen 6 3600, SPC Fortis 3 v2 cooling, Grizzly paste. HW Monitor shows me in idle around 32C, but when I fire up even a browser, the spikes start. Within 1-2 seconds the temperature can jump to 50C, drop to 40C, in a moment 30K, again 40+, 50+, 30.... etc. It seems physically impossible to me that the temperature can jump +/- 10-20C in a matter of seconds. Is this possible or more of a sensor or app error? The fan is consistently spinning at 670-770 rpm.

WHiNFO64 at idle. OCCT after 10 minutes of testing. Max 73.8C, but almost all the time the temperature stood at 72C. The fan sped up to 1400 rpm. After turning off OCCT the temperatures dropped by more than 20C literally in 3 seconds...
So I wonder.... Maybe I gave too much paste? It slightly squeezed out from under the heatsink. Not enough to spill all over the board, but a slight "overhang" formed around the CPU. Tightened up decently, more is not possible.


.

Well, so I took off a little paste. And actually quite a bit - this time I just gave a small "grain" to the middle of the CPU. Temperatures are even worse; in idle they hit 40C, in OCCT up to 74C....

Completely nothing has changed. Only the temperature increased by 1-2C on average, but it continues to jump whenever I load the CPU even for a moment (for example, opening a browser). It is possible that this time I gave a little too little paste, because I don't have much left.... I'll still try tomorrow with the paste added to Fortis - this time in the right amount.
But I don't know if this can be the main cause of the problems.... Somehow I doubt it.

With these MSI MAXes you have to be careful because they seem to have inflated voltages for the CPU in AUTO mode.

MOrtax max I have I don't know what the voltage is, but in AUTO it is certainly inflated - I do not know whether as much as indicated by hwinfo that 1.55V, but it is quite a lot and this voltage has the greatest impact on the heating even when you do nothing.

Paste should be given rather not too thin - I do not know how this your Fortris 3 but Grandis has a convex base so in the middle it presses and after the countries break. Although I don't worry about it.

I wouldn't worry about these jumps. Check how much HWINfo shows the TDP for the VRM - preferably screen with PRIME95 running in the background + fire up the software from MSI for OC under windows and set the test (With prime running in the background) lower voltage and not set to AUTO. Test until it doesn't get rash.

Generally OC control with these MSI from under windows as far as my eye is a failure. If I don't manage to add the CBS tab to BIOS then OC from under windows is the biggest failure I've encountered, really GIGabyte has CBS, ASUS was able to do it better too and MSI didn't give CBS rally there's no way to set p-stats just force fixed voltage. What's funnier changing the voltage from its own to deafult causes a BSODA...

Screens after applying OC yesterday and undoing it (so much that the voltage has to be manually lowered) - no longer CPU idls. Voltages so roughly set of course.
.

Hmm... At my place the voltages are currently as shown below. By default only the first position was on Auto. Then the voltage on the core jumped from 1.0xx to 1.4xx and with it the temperature jumped. After changing it to 1.225 it looks ok I think? Screen shots from HWiNFO before the voltage change and after the change. In idle the temperature does not exceed 40C (usually 36-38C). After 10 minutes of OCCT max 62C, average 60-61C. Acceptable
Till now I have not yet given new paste. I'll replace it when I buy Grizzly.

Apparently, there are no stupid questions.... But lowering the voltage on the CPU will not cause a drop in its performance?
.

No, lowering the voltage will not cause a drop in performance.

What about temperature spikes - they are normal, it's just that the chip (processor) itself is a tiny chip, covered with a large metal "cap" - it is impossible to take the temperature immediately from this small chip, so the spikes are also a normal matter.

Tomequ123 wrote:

That's why it's important to set the Load Line Calibration properly - level three, four

Any tips on how to go about it? Because this is completely virgin territory for me.

Unless it is held down by temperature and/or TDP limits - then it may be the opposite (lower voltage = willingly and longer holding a slightly higher clock).
Without playing around with a specific set of processor+board+memory, it's hard to say anything.

With this voltage boosting on Ryzen 3xxx - after all, the vast majority of boards do it, jumps to ~1.5V are nothing strange and this is normal behavior.
If you set the voltage low, you often won't be able to get as many MHz (several threads), to how much the factory turbo clock is able to boost....

rzymo wrote:

With this voltage boosting on Ryzen 3xxx - after all, the vast majority of boards do it, jumps to ~1.5V are nothing strange and it is normal behavior.

Suggesting to leave the voltage on the core on Auto and let it jump? Personally, these jumps don't bother me. I was more puzzled by the temperature. Although I have the impression that these temperature readings are not actual readings, but values calculated on the fly depending on the voltages (although I'm probably wrong as I know life ).

Basically, yes - Ryzen 3600 / 3600x by itself raises clocks so nicely that to achieve a little more you need to spend many hours on stability tests. And anyway it will be little more, for example, +100 MHz, whether there is a point - in my opinion, no.

We are coasting too far. I am not concerned with OC. I'm aiming for the hardware to work properly and stably, to use its potential, but not at 110%, and within reason (without forced OC). I started the topic because I was puzzled by the temperatures on the core. Both voltage and temperature spikes seemed disturbing to me. But if this is the way it should be and these spikes are normal phenomena, then there was no issue - I will restore the default settings. However, if it is more advantageous to stay with more stable voltages and much lower temperatures (even at the expense of a small loss of performance) then I can also leave it that way. So now I'm mainly asking for advice: option A or B.