I'm having the exact same issue, with virtually the same setup.
MSI z370-a pro MB
MSI Geforce 1070
fan cooled CPU--from Intel.
Nothing is OC.
Idle or just internet browsing the temperatures are in the 30's C, but under load while gaming it will run in the mid 70's C with spikes in the up 90's C. I just bought the computer, and I'm trying to make certain Intel thinks this is acceptable for the chips longevity. Those temp spikes seem very high.
...Then you are definitely overclocking at least a portion of the processor and the potential for higher temperatures is there.
In desktop processors, intel defines three important thermal thresholds:
- Ttrip, the Thermal Trip Temperature, is the temperature threshold at which the Thermal Control Circuit (TCC) is going to power off (thermtrip) the processor to protect the silicon from permanent damage as a result of a runaway temperature condition. Since it varies from one individual processor to another, Intel does not identify exactly where this threshold is. Suffice it to say, it is some amount (some say ~20c) above Tjmax.
- Tjmax, the Maximum Junction Temperature, is the temperature threshold at which the TCC is going to begin throttling processor performance in order to protect the processor from unsafe temperatures. This threshold also varies from one individual processor to another, but it is usually in a fairly tight range around 100c. Each of the processor's Digital Thermal Sensors (DTS) tell you digitally how far (as an offset) below the Tjmax threshold the local silicon temperature actually is. The highest possible reading, (an offset of) 0, indicates that the temperature is at or above Tjmax. That is, temperatures above the Tjmax threshold cannot be digitally represented. If, for example, a processor's Tjmax is 100c and the current temperature at a DTS is 105c, the DTS will still return (offset) 0, indicating that the temperature (still) appears to be 100c.
- Tcontrol, the Control Temperature, is the temperature threshold below which the motherboard's fan speed control subsystem is supposed to maintain the temperature. While temperature spikes above the threshold are allowed (subject to the processor's thermal load line), they should be of very short duration. Now, by definition, in the absence of specific information -- like the thermal load line of the processor present and the exact thermal dissipation characteristics of the processor's cooling solution -- which would allow the system's fan speed control subsystem to determine when the temperature should be allowed to spike above Tcontrol, the subsystem is required to have the processor cooling fan(s) running at full speed (100% duty cycle). Unfortunately, too many people do not understand this and they set operational curves for the fan(s) that do not meet this requirement. Worse, some manufacturers will cheat on (or just ignore) the setting so that they can claim that their system runs quieter (at the expense of the processor lifetime).
[Aside: As I said, you cannot determine the exact temperature at which the processor will thermtrip, but an individual processor's Tjmax and Tcontrol thresholds can be read from a specific Model-Specific Register (MSR). The IA32 Temperature Target MSR (0x1A2) provides the Tjmax threshold in bits 16-23 and the Tcontrol Offset in bits 8-15. Subtracting the Tcontrol Offset from the Tjmax threshold will tell you the value of the Tcontrol threshold. Tools like RWEverything allow you to read your processor's MSRs.]
So, what to do with this information?
- Adjust the configuration of your system's fan speed control subsystem so that the processor cooling fan(s) operate at full speed (100% duty cycle) at and above the Tcontrol threshold. Whether you implement this by anchoring the top of the fan speed control line/curve at the Tcontrol threshold or you implement an override that takes the fan to 100% if this threshold is exceeded is completely up to you. In my opinion, the latter method could result in ugly acoustics as the fan can rapidly speed up and slow down as temperature spikes occur.
- In more-sophisticated fan speed control subsystems, support is provided for smoothing fan response. Using this capability, you can ensure that the fan(s) do not over-respond to temperature spikes. Determining the best smoothing interval is very important; make it too long and longer duration temperature spikes may occur; make it too short and fan acoustics will suffer (and could become annoying as some folks have seen).
Hope this helps,
Thank you for your suggestions and help but ı strongly believe this 8th gen cpu's and 7700 k have serious problems about heating. İ just checked some forums tons of people have those heat spikes problem. İ have never faced this kind of a problem before even in my old amd and intel processors. Even i open calculator this god damn cpu create heat spike and goes from 30 degree to 55 degree and drops again.
This doesn't have to be a hardware issue with the cooling solution; it could be just the configuration of the fan speed control subsystem. The fan speed control configuration (likely configured in the BIOS but some boards come with runtime programs to do this as well) should be ensuring that, if the temperature of the processor is above its Tcontrol threshold, the fan/blower/pump should be operating at 100% duty cycle (i.e. full speed). Further, below the Tcontrol threshold, the duty cycle of the fan/blower/pump should be increasing from some minimum to 100% over some specific temperature range bounded at the top by Tcontrol. For example, from a minimum duty cycle of 20%, the duty cycle could be increased 4% for each degree that the temperature is above Tcontrol-20.
Hope this helps,
I have just read the very long temperature thread regarding the 7700 CPU.
I have just built a new PC.
AZUS STRIX Z370-H motherboard
Dominator 3200 GHz RAM, 16 GB.
MSI GTX 970 x 2 (SLI)
Kraken x62 water cooler.
WIN 10 pro
The temperature is all over the place when I game. Load seldom above 70 %. Spikes up to 85 degrees. Repeated warnings about temp from my software. The fans and pump keep the water temp at about 40 degrees C. I ran the diagnostic tool. Passed, and I saw no spikes when the test ran. Even at 100 % the temp was stable.
This can’t be right? Or healty?
Scott, you make valid points, but you seem to completely disregard any problems with the CPU itself. The 7700 thread suggests excatly that. These are expensive CPUs and if there is a problem that may kill it earlier than normal, I would like to send it back.
Unless Intel can find a solution to the problem.
If your spikes are not going above 85c, then everything is likely working as it should. I can't say for sure, since I do not know what your processor's actual Tcontrol temperature is, but it is typically somewhere around 85c. As I said earlier, your processor can run at its Tcontrol temperature constantly, even for its full warranted lifetime, without suffering any silicon degradation. Short duration spikes above Tcontrol similarly will not cause any degradation, though longer ones that take the temperature above the processor's thermal load line can have a longer term negative impact.
I regularly see folks expressing concern regarding temperature spikes. They are not an indication of any problem. Because Intel chose to use TIM, rather than solder, between the silicon and the heat spreader, short-term spikes will occur because the TIM doesn't dissipate this heat quickly enough. Provided your cooling solution is adequately dissipating heat and meeting fan speed control requirements (i.e. running fan/blower/pump at full speed when temperature is at or above the Tcontrol temperature), the duration of the spikes will be kept to a minimum and no issues will occur.
Where you need to be concerned is if the spikes go significantly above your processor's Tcontrol temperature and/or last for longer durations. This is not an indication of a problem processor (as some regularly assume); it is an indication of an inadequate cooling solution or an inadequate fan speed control configuration.
Folks have also expressed concerns regarding how busy the processor seems to be. Its not the processor's fault. I blame this on Windows 10, which has a lot more going on in the background (my facetious response: it's all that spyware that is collecting all sorts of information about you and sending it back to Microsoft). You'll note thst it's only Windows 10 users who express this complaint ;^)
Now I found some solution to my case.
1 - I delited my processor, the thermal solution from intel is terrible, this gave me - 23,0º celsius degree.
2 - XMP profile to overclocked RAM is doing overvoltage in every aspect, so you need to undervolge the entire thing.
3 - use a good thermal solution and thermal grease, I'm using right now Thermal paste from Grizzly Bear, Kryonaut.
Now I can get at maximum 70º celsius degrees, before that it get 100º degrees instantly with prime95.
I'm very upset with intel and I'm thinking that will be my last intel computer.
I really want to know how a 95W TDP processor, running in stock configuration can instantly reaches 200W.