[SOLVED] Added new CPU, PC has problems. Help if you can?

[Lux89]

My old build was;
Intel Core i5-8600K LGA 1151 4,80 GHz OC
Corsair Hydro Series, H150i PRO, 360mm Radiator, Advanced RGB
MSI RTX 3070 Gaming X Trio 8GB
MSI Z370 Gaming M5 8Gen (last BIOS update)
Patriot Viper RGB LED, 16 GB Kit (2x8GB) 3000Mhz, black, PVR416G300C5K
Patriot Viper RGB LED, 16 GB Kit (2x8GB) 3000Mhz, black, PVR416G300C5K
SSD Intel 660p Serija 1TB NVMe M.2 disk, SSDPEKNW010T801
HDD Interni Seagate SSHD FireCuda 3.5" 2 TB, 7.200 rpm, ST2000DX002
SSD Samsung 500G 840 EVO
HDD 1TB and HDD 500GB 7.200PRM
Napajanje Corsair PSU 850W Professional Platinum Series HX850i, EU version
Corsair Obsidian Series 500D RGB SE Premium
Everything was OK with this combination, but needed more cores and threads. So I added a i9 9900K CPU. Then my PC started to have random freezing with the new CPU, always needed to restart - shutdown the PC to use it again. I tried a lot of things to fix this and nothing helped except a fresh copy of Windows 10.
So I fixed that and now I wanted to test the CPU before overclocking it. But it happens that the system restart itself when doing PRIME95 blend test. Don't know why.
Only thing I noticed and found out was this Critical Event; https://ibb.co/wwcg5W8
I don't know if it's maybe the PSU, the new CPU, some other problem with incompatible component with the new CPU, heat? Dunno...
Any ideas how to fix this or at least know what is causing this restart?

 

[Phaaze88]

A)VRM is either overheating or unable to keep up with the 9900K's power demands, and bails the heck out when things get rough.
B)You didn't clear CMOS after the cpu swap; some of the old cpu settings carried over to the new one, but it didn't agree with them.
C)The mixed ram kit isn't actually stable. 8 passes of Memtest86 were already done?
[You said you 'tried a lot of things' but weren't specific on it.]
Or perhaps a combination of the above.


Event ID 41 means Windows failed to shut down properly. The cause could be literally anything.

 

[dorsai]

How's your case airflow ? Is the VRM area getting plenty of air flow ? Typically AIO coolers result in decreased airflow on the VRM vs a normal heatsink fan combo...you may need to dedicate a fan blowing onto the VRM to keep it happy at max temps.

 

[Lux89]

A)VRM is either overheating or unable to keep up with the 9900K's power demands, and bails the heck out when things get rough.
B)You didn't clear CMOS after the cpu swap; some of the old cpu settings carried over to the new one, but it didn't agree with them.
C)The mixed ram kit isn't actually stable. 8 passes of Memtest86 were already done?
[You said you 'tried a lot of things' but weren't specific on it.]
Or perhaps a combination of the above.


Event ID 41 means Windows failed to shut down properly. The cause could be literally anything.

A) I use HWiNFO64 and didn't notice any problems with temps there. All was in solid/decent temp zones. No red at all or even close, at least to my knowledge. For the power demand wouldn't the Z390 be able to do it? I mean, the motherboard is not a cheap one: https://www.msi.com/Motherboard/z370-gaming-m5.html (Where can I check those stats/things to be 100% sure?

B) Nope I didn't, only thing I did in BIOS was reset settings back to default. I didn't remove the battery and place it back to reset CMOS. Could try this.

C) When do you mean mixed, it's the same type and models just 2 kits of 2. And when you say 8 Passes you mean to run the test 8 times or just the regular test to do his own? I did the test 2x and there was no problem detected.

 

[Lux89]

How's your case airflow ? Is the VRM area getting plenty of air flow ? Typically AIO coolers result in decreased airflow on the VRM vs a normal heatsink fan combo...you may need to dedicate a fan blowing onto the VRM to keep it happy at max temps.

Well the case itself has really good air flow in my opinion and I clean it each 2 weeks with high pressure air, so my PC is spotless when it comes to dust etc; https://postimg.cc/2VdYBJ9J
The front has a push pull combo and 6 Corasir fans, it's a lot of air flow. I will need to check HWiNFO64 and see if it has those stats, temps for VRM.

 

[Phaaze88]

A) I use HWiNFO64 and didn't notice any problems with temps there. All was in solid/decent temp zones. No red at all or even close, at least to my knowledge. For the power demand wouldn't the Z390 be able to do it? I mean, the motherboard is not a cheap one: https://www.msi.com/Motherboard/z370-gaming-m5.html (Where can I check those stats/things to be 100% sure?

Some Z390s don't even handle the 9900K well. If you're the type to prefer running without power limits, overclocking, or high performance power plan, things get worse for the less than adequate boards.
It's like AMD's FX 8000 and 9000. Mobo vendors listed these cpus as supported(as far as bios micro code goes) on all their AM3+ boards, but some of the boards actually handling them was a different matter.
Some of the weaker VRM boards would overheat or crash due to the high power draw they weren't designed to handle.

Don't get me wrong, some of the Z370s can handle a 9900K without issue... but then there's others that just can't, or require the user to tune settings down a bit.
If hwinfo shows the VRM isn't thermal throttling, that's good.
Since the system completely flips a table upon running Prime95, that's bad. But Blend mode runs ALL the other test options, one at a time, and it runs each one for 10-20mins straight before moving to the next one.
I don't know which test it starts off with, and they all apply different loads to the cpu/cache/ram.


Something that I just recalled: since your board is well equipped to handle an 8700K, and the 9900K is in another league from it in terms of power and current draw, it could just be that the motherboard's cpu current limit(default is 100%) is too low.
Try raising that(there are multiple settings), though this would lead to higher VRM temperatures in exchange. You should be able to find it in advanced cpu settings relative to power.

B) Nope I didn't, only thing I did in BIOS was reset settings back to default. I didn't remove the battery and place it back to reset CMOS. Could try this.

Do give that a go then and see if it still crashes.

C) When do you mean mixed, it's the same type and models just 2 kits of 2. And when you say 8 Passes you mean to run the test 8 times or just the regular test to do his own? I did the test 2x and there was no problem detected.

Same type and model = higher chance of being compatible. It does not guarantee compatibility, thus the user has to do the testing and certification that the manufacturer normally does with the original kits.

Memtest86 by default does 4 passes, or laps, of 13 different tests. The latest version allows for more passes.
The more ram you have, the longer the testing takes. 32GBs and 8 passes takes like 13-14hours last I checked - I slept through most of it. A small price to pay to safeguard against OS corruption.

 

[Lux89]

Same type and model = higher chance of being compatible. It does not guarantee compatibility, thus the user has to do the testing and certification that the manufacturer normally does with the original kits.

Memtest86 by default does 4 passes, or laps, of 13 different tests. The latest version allows for more passes.
The more ram you have, the longer the testing takes. 32GBs and 8 passes takes like 13-14hours last I checked - I slept through most of it. A small price to pay to safeguard against OS corruption.

I was wrong, did some different test but not the Memtest86, so I downloaded the last free version and did the full test, took 5;30h for 32GB RAM. And it did FAIL! Here is the screenshots;

1. https://postimg.cc/mPwk9PzM
2. https://postimg.cc/2bPSYvMp
3. https://postimg.cc/gwjz6NPF

So what exactly does this mean now? My memory is dying, it's not compatible, something other? Since all of this did work with the 8600K.

 

[Phaaze88]

I was wrong, did some different test but not the Memtest86, so I downloaded the last free version and did the full test, took 5;30h for 32GB RAM. And it did FAIL! Here is the screenshots;

1. https://postimg.cc/mPwk9PzM
2. https://postimg.cc/2bPSYvMp
3. https://postimg.cc/gwjz6NPF

So what exactly does this mean now? My memory is dying, it's not compatible, something other?

It doesn't mean too much - not yet, but the test should pass with NO ERRORS. If freezing/crashing occurs during the test, that's also a bad sign.
Was this run done with XMP enabled or disabled?
If XMP was enabled, then it's possible that it can be made stable with some more voltage.
If XMP was disabled, then it's highly likely the 2 kits aren't compatible at all.

Since all of this did work with the 8600K.

You could've just gotten lucky - it takes time for memory errors to screw your OS up - but was Memtest86 ever run with your 8600K like it was just recently?

 

[Lux89]

It doesn't mean too much - not yet, but the test should pass with NO ERRORS. If freezing/crashing occurs during the test, that's also a bad sign.
Was this run done with XMP enabled or disabled?
If XMP was enabled, then it's possible that it can be made stable with some more voltage.
If XMP was disabled, then it's highly likely the 2 kits aren't compatible at all.


You could've just gotten lucky - it takes time for memory errors to screw your OS up - but was Memtest86 ever run with your 8600K like it was just recently?

No freezing/crashing during the test. Yes XMP was turned on, I just use the automatic button for that in the BIOS.

I am not 100% sure but 90% that I did the exact test with the old setup and had 0 errors. But can't be 100% sure... What is my next step, what do to now?

So for now we are not even close to exactly know why the PC is restarting on Prime95 stress testing, any other steps to narrow it all down and find the reason and fix it? I got one more PC different ram but also DDR4, so could use those or? I also didn't find any stat for VRM heat in HWiNFO64 :/

 

[Lux89]

Tried to get a dump file to check what is happening but no luck, so far it's not making one or I can't find the file itself Oo Need to work a bit more on it since I was reading a lot of similar problems and people always asked for those dump files how they can point specificly what the reason was for the critical error - restart.
I noticed I can force the restart the PC with Prime95 in a few minutes or seconds doing the small or smallest FFTs stress test.

 

[Phaaze88]

I am not 100% sure but 90% that I did the exact test with the old setup and had 0 errors. But can't be 100% sure... What is my next step, what do to now?

Do it again, but with XMP off. If that passes, then it means the cpu's Internal Memory Controller needs more voltage to run XMP.
If that also fails, then it means you drew the short straw on IMC quality.

I also didn't find any stat for VRM heat in HWiNFO64 :/

There is one, but it's usually disabled - or maybe it's just my board...
When you look through hwinfo, is there a red circle with an X in it, followed by Msi EC(Embedded Controller)? Right-clicking and enabling monitoring should allow you to view it.
On my end, it's disabled by default, because it can cause latency, inaccurate readings, or something like that. Should be ok to look at it now and then - don't leave it on all the time.

I noticed I can force the restart the PC with Prime95 in a few minutes or seconds doing the small or smallest FFTs stress test.

Small and smallest FFT make the cpu go big on power draw, unlike medium and large FFT.
Cpu is tripping the current limit, or the VRM is throttling. Most likely the former is happening, because if the VRM is thermal throttling, it'll just force the cpu to drop clocks.
If it hits critical limit, then they'll shut down. Now, it takes time for the VRM to heat up like that, unless their cooling is crap.


Yikes, I'm starting to go in circles.
Did you ever increase the cpu current limit? Go into bios > Advanced Mode > OC > DigitALL Power > Cpu Over Current Protection. Auto is usually 100%.

 

[Lux89]

Do it again, but with XMP off. If that passes, then it means the cpu's Internal Memory Controller needs more voltage to run XMP.
If that also fails, then it means you drew the short straw on IMC quality.


There is one, but it's usually disabled - or maybe it's just my board...
When you look through hwinfo, is there a red circle with an X in it, followed by Msi EC(Embedded Controller)? Right-clicking and enabling monitoring should allow you to view it.
On my end, it's disabled by default, because it can cause latency, inaccurate readings, or something like that. Should be ok to look at it now and then - don't leave it on all the time.


Small and smallest FFT make the cpu go big on power draw, unlike medium and large FFT.
Cpu is tripping the current limit, or the VRM is throttling. Most likely the former is happening, because if the VRM is thermal throttling, it'll just force the cpu to drop clocks.
If it hits critical limit, then they'll shut down. Now, it takes time for the VRM to heat up like that, unless their cooling is crap.


Yikes, I'm starting to go in circles.
Did you ever increase the cpu current limit? Go into bios > Advanced Mode > OC > DigitALL Power > Cpu Over Current Protection. Auto is usually 100%.

Just to be clear you want me to put back the old 8600K and then test with the XMP off, or do you want me to test the new 9900K with the XMP off? Wasn't sure, so better to double check this part. IMC quality?

Nope, don't have it in HWiNFO64, only my water cooling sensors are disabled. Was reading somewhere how many of the MSI don't have that sensor on the board itself. So I guess that is the case here. What about MOSFET temps, got that one.

Nope didn't play with the current limit, any I did rest BIOS to deafult and CMOS reset also so everything is 100% clear there. Only turned on XMP. Should I increase Cpu Over Current Protection or stay or 100% limit?

 

[Phaaze88]

Just to be clear you want me to put back the old 8600K and then test with the XMP off, or do you want me to test the new 9900K with the XMP off?

9900K. XMP off.
Just like how core quality determines how far one can push core overclocks, the quality of the memory controller determines how far one can push memory overclocks.
Higher frequency and number of dimms in use increase difficulty. Yes, 4 sticks at 3000mhz is harder on the controller than 2 at 3000mhz, though 3000mhz isn't particularly hard to run in this day and age...

Nope, don't have it in HWiNFO64, only my water cooling sensors are disabled. Was reading somewhere how many of the MSI don't have that sensor on the board itself. So I guess that is the case here. What about MOSFET temps, got that one.

Ok, it's a little different.
MOSFETs should be good up until 90C. They're rated for higher than that - like, 120C - but still, you don't want them getting THAT high.

Nope didn't play with the current limit, any I did rest BIOS to deafult and CMOS reset also so everything is 100% clear there. Only turned on XMP. Should I increase Cpu Over Current Protection or stay or 100% limit?

It's best to do things one at a time. Which do you want to work out first? The memory stability, or the restarts/shut offs?

 

[Lux89]

9900K. XMP off.
Just like how core quality determines how far one can push core overclocks, the quality of the memory controller determines how far one can push memory overclocks.
Higher frequency and number of dimms in use increase difficulty. Yes, 4 sticks at 3000mhz is harder on the controller than 2 at 3000mhz, though 3000mhz isn't particularly hard to run in this day and age...


Ok, it's a little different.
MOSFETs should be good up until 90C. They're rated for higher than that - like, 120C - but still, you don't want them getting THAT high.


It's best to do things one at a time. Which do you want to work out first? The memory stability, or the restarts/shut offs?

OK, will test the 9900K with XMP off today and tell you what results I got with that.

Well I guess the reason for the restart/shut off since the stability is #1 here. For example with the OLD CPU I was mining stuff sideways when not gaming. And my PC was turned on months with no shut down or rest, my PC was super stable. Now I upgraded the CPU for some rendering and plotting work I need to do when not gaming and for that I also need a stable system. Since if it shutdowns or restarts when doing that I lost everything. So I am look to have my system stable and then rise the power of the CPU as much as I can when OC. But as you can see, can't even have it stable on default settings. So yeah, that would be the #1 thing to fix.

 

[Lux89]

Did the new Memtest86 with XMP off., 0 errors this time.

 

[Phaaze88]

Ok. Then time to try upping the voltage with XMP.
Manually enter the XMP frequency, PRIMARY timings(should be on the sticker on the back of the sticks), and instead of the 1.35v, enter 1.40v instead.
Test.

By the way, what are the VCCIO and VCCSA values?

 

[Lux89]

Ok. Then time to try upping the voltage with XMP.
Manually enter the XMP frequency, PRIMARY timings(should be on the sticker on the back of the sticks), and instead of the 1.35v, enter 1.40v instead.
Test.

By the way, what are the VCCIO and VCCSA values?

For now I really don't care much about XMP and RAM speed, since my PC will still restart on Prime95. With XMP off or on. So that dose not help much. I literally turned on Prime95 today and as I pressed the button to stress test the PC restarted, like really WTF? It didn't have time to heat up or anything, the same moment I pressed to run the test BOOM restart Oo

And I am still not any closer to find the reason for it all. I mean if it's the new CPU I would return it to the shop and ask for the new one, but since I don't know what, why or how. Nothing that I can do in this moment.

Sorry but never heard of VCCIO and VCCSA, those are settings for what, the CPU? Need to see where I can find those.
Thanks for the help for now, but this is really stressing me out since I can't seem to fix it or at least find the reason for it and be 100% sure it's that...

 

[Phaaze88]

For now I really don't care much about XMP and RAM speed, since my PC will still restart on Prime95. With XMP off or on. So that dose not help much. I literally turned on Prime95 today and ass I pressed the button to stress test the PC restarted, like really WTF? It didn't have time to heat up or anything, the same moment I pressed to run the test BOOM restart Oo

Did you raise Cpu Over Current Protection limit yet?
My X299 board only has 2 measly settings higher than 100%: 140 and 200%. This is bloody silly on my end, because the 7820X trips the 140% current limit and restarts the PC without much effort, so when overclocking, I have to put it at 200%.