My ecores... are causing bluescreens?

[iameatingjam]

Okay I'll try my best to keep this short, I know I have a tendency to ramble... My 14700kf was working great for about a month. Then a couple days ago I started getting bluescreens, bluescreens, bluescreens constantly no matter what I was doing. Even when booting up from USB to reinstall windows, bluescreen. This was the most repeatable bluescreen, I press 'install now', it bsods exactly 3 seconds later, before even getting to the part where I can partition drives. They are either 'kmode exception not handled' or 'machine check exception'


So I tried just about everything I could think of

Updated the bios
cleared the cmos
loaded optimized defaults
turned off all overclocking, undervolting, xmp etc
reseated the cpu and cooler
tried using a different power supply
tried running each stick of memory individually, and ran them both through memtest86, at least that didn't bluescreen lol, and it was a pass.
Disconnected all but one ssd (and tried swapping them) and disconnected non essential things like usb expansions and printer, webcam etc.


And nothing made any difference. At this point I was feeling pretty confident it was either the cpu or the motherboard so I ordered a 12100 to find out which. if the 12100 works, then its the cpu. If it doesn't work, then its the motherboard.

But then I started thinking about the differences between a 12100 and a 14700k why one might work over the other etc and long story short I ended up figuring out its the ecores that are causing the problems. Ecores disabled fixes everything. I was finally able to do the fresh install of windows, linux works, I can even run cinebench, there doesn't appear to be any limit as far as how much power is delivered to the cpu, its the same as before ~260w.

So I started turning the ecores back on one at a time, they worked up until core 5. So something is up with that second cluster of ecores ( or whatever interfaces with it). Good news is, I now have a working computer. Bad news is, that 12100 probably aint going to help me now.

And now I don't have a clear plan of action, how do I know if its the cpu or mobo? (Apart from buying some very expensive parts for testing) Anybody have any idea... what might give me a clue here?

Wishful thinking has me thinking cpu but thats probably because I don't want to deal with a gigabyte rma, install a new motherboard, and because with the 12100 at least I'll have something while waiting for a cpu rma. Though I realize statistically, mobo is more likely to fail. I did try examining the pins and nothing stuck out at me, but its pretty hard to see.

TLDR: 14700kf works for a month then gives constant bsods. Disabling ecores fixes the problem, and with up to 4 ecores enabled everything works perfectly even under max load. I enable the fifth ecore, instant blue screens even with light tasks. More indicative of cpu or mobo problem? Something else?

And sorry I never know where to put these cpu troubleshooting threads. Would system builders advice have been better?

 

[Dr. Dro]

It's looking like it's a CPU RMA situation, man. Install the 12100 and file an RMA request with Intel, IMHO. You paid for a new CPU that has 12 E-cores and 12 E-cores you shall get. You already went enabling them one by one, so if the memory tests pass, and you get sent back to BSOD hell by enabling one given core, then it seems like you need that CPU exchanged ASAP.

Don't think it's your motherboard mate. Seems way too specific.

 

[iameatingjam]

It's looking like it's a CPU RMA situation, man. Install the 12100 and file an RMA request with Intel, IMHO. You paid for a new CPU that has 12 E-cores and 12 E-cores you shall get. You already went enabling them one by one, so if the memory tests pass, and you get sent back to BSOD hell by enabling one given core, then it seems like you need that CPU exchanged ASAP.

Don't think it's your motherboard mate. Seems way too specific.

Thank you. Thats what I was hoping to hear.

 

[ir_cow]

If you disable the e-cores and all is good. Either you have a cooling problem or a defective CPU. I would lean towards a CPU issue.

 

[Solaris17]

Thank you. Thats what I was hoping to hear.

Iv rmad with intel before it was an absolutely painless experience if you can’t go through your original purchaser

 

[phanbuey]

usually indicative of not enough voltage to the e cores....

 

[ir_cow]

usually indicative of not enough voltage to the e cores....

This is true. If it's completely stock, that is a problem, but if it's a undervolting issue... Yeah that's self inflicted.

 

[ShrimpBrime]

Got a buddy you can send the cpu to test in a different board? I'd do it for you, but you don't know me and I don't know you to accomplish this type of favor.

GL!!!

 

[iameatingjam]

This is true. If it's completely stock, that is a problem, but if it's a undervolting issue... Yeah that's self inflicted.

Do you mean if there was ever an undervolt or if there is one currently? Because everything is stock now. But there was an aggressive vcore offset before.

 

[ir_cow]

Do you mean if there was ever an undervolt or if there is one currently? Because everything is stock now. But there was an aggressive vcore offset before.

If the offset caused the bluescreen and now its gone, that means you caused it with the offset being to low.

 

[iameatingjam]

If the offset caused the bluescreen and now its gone, that means you caused it with the offset being to low.

I see. So an undervolt can do permanent damage? That is not what I thought was the case... Well I guess I wont be undervolting my replacement (if I can get one).

It did help in cinebench but thats not real world use for me anyway. Might have to put up with higher fan rpm but, first world problems I guess

 

[InVasMani]

You probably simply dropped voltage lower than expected for operation by setting it too low with offset and/or LLC. It could be throttling as well, but that shouldn't really cause a crash just reduced system performance.

 

[ratirt]

So an undervolt can do permanent damage, good to know, thanks.

Udervolt will not cause damage to the CPU. Only overvolting can do that. I think @ir_cow meant, you cased the instability and the blue screens by undervolting. Which happens very often when you try to lower temps and you crank down CPU's voltage.
It was self inflicted instability rather than CPU's fault.

 

[iameatingjam]

Udervolt will not cause damage to the CPU. Only overvolting can do that. I think @ir_cow meant, you cased the instability and the blue screens by undervolting. Which happens very often when you try to lower temps and you crank down CPU's voltage.
It was self inflicted instability rather than CPU's fault.

I see thats what I had originally thought was the case. Thing is, the instability ( at least with that second cluster of ecores) continued even after the undervolt was removed. At first I thought filesystem corruption. But when it happened even booted from usb I knew that wasn't the case.

I don't even know if undervolting had anything to do with the blue screens really. The offset was in place since day one, seemingly stable then one day weeks later suddenly blue screens galore, undervolt or no. So who knows?

But either way I think I'm going to be more careful going forward. If I undervolt it will be a small one. Think I'll dial back my ram oc too. Stable PC > 1-2% faster PC.

 

[ratirt]

I see thats what I had originally thought was the case. Thing is, the instability ( at least with that second cluster of ecores) continued even after the undervolt was removed. At first I thought filesystem corruption. But when it happened even booted from usb I knew that wasn't the case.

I don't even know if undervolting had anything to do with the blue screens really. The offset was in place since day one, seemingly stable then one day weeks later suddenly blue screens galore, undervolt or no. So who knows?

But either way I think I'm going to be more careful going forward. If I undervolt it will be a small one. Think I'll dial back my ram oc too. Stable PC > 1-2% faster PC.

I remember my UC and OC and i can tell you now. Sometimes when you Undervolt too much and the process causes system instability, even if you revert everything, it will still be unstable. Reinstalling windows is what you can do. Use Stock settings for everything and reinstall windows. If that doesn't help, maybe the CPU has an issue and you would need to RMA it.

With UV you cant damage your hardware but you can make Windows unstable. If you want to UV, sure do it but decrease the voltage in small steps and check stability after every decrease with Cinebench for instance.

 

[Calenhad]

RMA the processor. You have done your due dilligence and more.

 

[iameatingjam]

I remember my UC and OC and i can tell you now. Sometimes when you Undervolt too much and the process causes system instability, even if you revert everything, it will still be unstable. Reinstalling windows is what you can do. Use Stock settings for everything and reinstall windows. If that doesn't help, maybe the CPU has an issue and you would need to RMA it.

With UV you cant damage your hardware but you can make Windows unstable. If you want to UV, sure do it but decrease the voltage in small steps and check stability after every decrease with Cinebench for instance.

Yeah thats why I booted off usb to try and reinstall windows, but I couldn't even do that, with everything at stock. It blue screened before I could even get to the partition part. And the usb drive wouldn't have filesystem damage after the undervolt, I made it on my laptop. And it wasn't even connected to my desktop until after undervolt was turned off so clearly something else was up. I couldn't get windows reinstalled until the ecores were turned off.

I might have got carried away with undervolting but I went slow and ran cinebench after every bump. Scores kept going up and temp kept going down, and it seemed stable at the time. It did twice what my 13600k would and 4 times what my 11600k would do before acting weird. This cpu would just keep going and going I had never experienced anything like that. But also this is the only cpu that started having defective cores after a month. So, related? Idk.

And yeah I'm going to RMA it. As soon as my 12100 arrives.

 

[ir_cow]

I see. So an undervolt can do permanent damage,

Undervolt didn't do permanent damage.

 

[iameatingjam]

Undervolt didn't do permanent damage.

Sorry I'm confused. You said the ecore thing was self inflicted. How could that be if undervolting is off and it's a fresh install of windows? Just a miscommunication I guess.

 

[FoulOnWhite]

Sorry I'm confused. You said the ecore thing was self inflicted. How could that be if undervolting is off and it's a fresh install of windows? Just a miscommunication I guess.

He means, sometimes too much of an undervolt can cause a problem, which would be self inflicted, in this case, the CPU is faulty.

 

[ShrimpBrime]

Nobody can say running the cpu out of spec couldn't damage it. If it causes any type of instability, of course there's risk of damages. Under-volt or other.

Cpu needs to be dropped into another board for testing. I've got 3 1700 boards, 2 for clocking one for testing. Current 13700K DDR5 mem controller is toasted to single channel. Had to aquire a second board to test the cpu. Cpu is bad. It's how it goes. Money well spent, I can upgrade one of the kids rigs. Then test with that when needed. I agree though, I think cpu is bad. You could try adding voltage and see if you just made a vcore hungry cpu. I've had a few get cold spoiled for sure. Seems once they've gotten cold with OV, the chip wants more even on ambient.

Besides, you clock Intel the v-core by raising it. Keep v-droop to a minimum with LLC. They just run hot all p and e core loads avx. So you OC to the throttle point of your cooling. If you can't hang at stock, gotta upgrade the cooling. It's the only way to run an Intel chip unfortunately. The E core thread chase to AMD has Intel reaching. They should have done 10P i5 and 12P I7/i9x and left the nano cores in tablets. [Opinion]

Good luck.

 

[ir_cow]

He means, sometimes too much of an undervolt can cause a problem, which would be self inflicted, in this case, the CPU is faulty.

I was to cryptic lol.

 

[FoulOnWhite]

They should have done 10P i5 and 12P I7/i9x and left the nano cores in tablets.

That would have been pretty nice.

 

[iameatingjam]

He means, sometimes too much of an undervolt can cause a problem, which would be self inflicted, in this case, the CPU is faulty.

Just while its active? Okay that makes sense. I guess I just read it differently, my understanding of the world was shattered there for a second. Glad everything is back in place lol.

RMA the processor. You have done your due dilligence and more.

Thanks. I'm just happy I have intel to work with and not gigabyte

And thanks for everyone else's input. I'll send out the rma once the 12100 comes in.