You said it was on adaptive mode correct?
I wonder if the freezes are because you are not getting enough power. I would try setting it to "Auto" and re-running the tests to see if stability improves.
As to the temps. Anything under 100C has been deemed OK by intel. As to the rapid temp increase/decrease : the sensors on these chips have been upgraded to receive a reading every 1/4 seconds and on top of that they are located on the DIE which will shift in temperature almost instantly due to increase/decreases in voltage. Also remember the DIE on these chips is very small so there is not much surface area to heat up or to transfer heat to your block.
All that being said I would try to put your voltage on AUTO and see what results you get from those same tests. If on auto you are still seeing issues I would report back. A Bios update wouldn't hurt either.
Hey TGrable , when i said adaptive mode i meant Auto mode , it's the default setting for voltage on my cpu. and i doubt it's not getting enough voltage since it usually drives it between 1.2-1.34 according to HWmonitor. though whats weird to me is that the CPUTIN sensor (which is apparently the CPU temp sensor from the motherboard shows slow increments every time it refreshes as i start running a benchmark and it locks around 70C after a while without moving , whereas on the very MILLISECOND I press to start a benchmark or a stress test , the CPU package sensor goes INSTANTLY to 90C (90C is an example of prime95 test)
Here's a video to demonstrate the occurrence : i7 7700K Unstable CPU Temp - YouTube
Watch in the video the CPUTIN Sensor that is the sensor of the CPU provided by the motherboard how it changes temperature in a much more logical way whereas the CPU package sensor goes 40 degress + and then 40 degrees - In 0 time.
I should also mention that because my fans are being controller by the CPU package sensor that means that they go extremely high on RPM when the benchmark is running but as soon as i stop it , it would go to a low RPM which doesn't make sense to me.
Now lets assume that the CPUTIN sensor is a good indicator and the CPU package one is faulty , that means that when my cpu temp rises slowly according to CPUTIN and settles around 70C , it stays there a long time after i finish the stress test just because the fans have already went back to low rpms and barely help cooling it down since the CPU package sensor is lying (at least that's what i think) and claims that the cpu is already cool.
BTW this kind of behavior is relevant to every single piece of software that's making my CPU work , not only Stress tests (prime95 or realbench)
Regarding the freezing issue I would like to gather the following information;
- What is your BIOS version?
- Can you share the results you got from running the Intel® Processor Diagnostic Tool?
- Have you ran these tests with the minimal configuration? Do you get the same freezing situation?
- Make sure you are running the latest chipset driver for your model, find it here;https://www.msi.com/Motherboard/support/Z270-GAMING-M3.html#down-driver&Win10 64
- I would also recommend re-applying your thermal grease.
The situation with the fan speed while running the benchmark is completely fine, remember that you are running a stress test which is going to stress the processor, and the processor will need the fan to cool down; therefore, the RPM will increase.
I would like to share with you that regarding the spiking situation on the Intel® Core™ i7-7700K Processor there has been an official statement which you can find here; https://communities.intel.com/message/471425#471425, post #549.
Hello Amy ,
My BIOS version is 1.2 and according to my Motherboard's manufacturer it is also the latest there is for my motherboard.
My results are written in this file : https://ufile.io/kek7k
There is no point in lowering those tests to minimal since what I'm trying to achieve is loading my CPU to stress test it and see how it reacts (Apparently looking at the occasional freezes it isn't very stable)
I have tried reapplying the Water block of my H100i and made sure it sits very well on top of my CPU yet still nothing happened.
I know the CPU temp makes my fans increase their RPM i set the fan curve to my liking though i'll be honest with you , It seems as if having my fans spinning really fast doesn't change anything about the CPU temp. I still feel like the reading is just false which is making the fans speed up for nothing.
I will add two more things :
I found out that my Vcore Voltage moves between 0.928V- 1.288V and my VID from 0.658 sometimes to 1.345V It seems also extremely unstable..
Also during the test of IPDT I noticed that no parts of the test made my Pc Freeze for a few seconds except for the CPU Load part , which makes sense considering the fact that many other CPU Stress tests made my computer freeze as well.
The statement you linked me to basically told me to not overclock , alright i got that and i didn't , everything's at its default settings but it did not mention anything about my problem , I'm saying my CPU temps change from 40C to 90C in less than a millisecond.. The moment i start Prime95 for example to load my CPU intentionally it would climb to 90C in NO TIME , You can watch the video i linked above to see what i mean.. This does not make sense to me.. Are you sure it's supposed to behave this way? as soon as i stop a process it just makes all the heat that was there a moment ago disappear right away? it's like my fans and cooling system's doing nothing at all cause on a click of a button all the heat disappears according to the sensor.
SleVeran, I would like to know if there is a chance to test different RAM? You can try a different brand that complies with the following DDR4-2133/2400, DDR3L-1333/1600 @ 1.35V. This for testing purposes because at this stage it seems like the replacement should be the next step.
Hello Amy , Thank you for your help
I do not have another piece of DDR4 Ram to check with though i will say that I ran windows Memory diagnostics and it said there were no error detected.
Next i figured that those freezes show instability of the processor so i figured maybe the voltage my processor is getting from AUTO mode on my motherboard is not good enough to run the intel boost clock of 4.5Ghz , So what i did is try to set my clock at 4.5Ghz by changing the multiplier to 45 and then run realbench , as long as I got freezes i would go back to the bios and feed it with a bit more voltage , i found out that 1.28 wasn't really stable so i bumped the voltage to 1.3V and i noticed that there were no more freezes ( i didn't test for a very long time though the freezes usually come in after a few seconds into the test and i waited a few minutes with no freezes in the horizon seen i felt quite good ). i changed the voltage back to adaptive mode and i assume that no longer is my chip unstable , I'm assuming for now it's just a problem with the motherboard Automatic voltage mode working with Intel Boost clock.
Can you approve that my conclusion is right and that if i don't get freezes now that means it was just a bad adjustment of voltage by my motherboard?
Thank you again for your help , i'll be awaiting reply.
I'll try to run a longer stress test soon and tell you if there were freezes on a longer period of time.
SleVeran, this information definitely changes this situation. On the default settings there should not be a freezing behavior, so based on this I agree with your conclusion. What are the chances of getting in touch with MSI and share this? It would be helpful if their support can comment on this voltage matter.
Hello Amy ,
First of all I want to thank you for your help figuring this out and helping me by giving guidelines and directions to solve the issue.
I do not want to say my conclusion is absolute for now , I'll find a time in the next couple of days to run a long stress test since a few minutes of a stress test can't guarantee the problem has been solved though it does look better for now.
I'll wait with this post before posting my final conclusion , I will try to be as quick as i can with it.
Thank you again Amy , I will come back as soon as I can get a longer test and watch the freezing problem after a while. Please leave the post open for comments as for now.