Cpu stopped working 3950

[Metalface Mark]

My pc stopped working today, unsure if it’s the board or the cpu. I’ve tried resetting the cpu a few times with no luck, and have taken it out to have a look.

I can’t remember what it looked like when I put it in, but there’s a little cluster of pins where it looks discoloured, but no idea if it’s normal or not, trying to workout whether to rma the cpu or the motherboard, cpu has been running at stock since I got it.

Any advice? Pc won’t post and gpu fans just run 100% rather than ramping down as it would with a Normal post, was playing RDR2 when it stopped working

 

[Metalface Mark]

Had plans to stream. So figured the extra cores would save running a separate pc but I haven’t got round to setting up the streaming stuff yet.

 

[Metalface Mark]

Maybe begin by actually telling the truth when asking a question ? It hasn't been running "at stock" since you got it, has it ? In fact, at least once it has had an all core clock of 4.25Ghz ( https://forums.overclockers.co.uk/posts/33243719 ) ( https://www.3dmark.com/3dm/42278420? ).
So, what Vcore did you use to get that and more importantly have you actually been running that all core clock ever since ?

Oh I played with overclocking for a few hours, but the benchmarks in game were actually slower.

Running anything other than auto volts put the temps up too high running cinebench, the highest voltage I had it at were 1.35, which is less than what it runs at on auto.

if you look at my profile on 3dmark you can see I did a stress test earlier today with everything at stock cpu speeds.

I actually forgot I had briefly overclocked it when I was trying to break 10k on port royale, however, nothing was done outside of the chips operating speeds and voltages.

 

[Metalface Mark]

What board did you put it in. Were the VRMs man enough?

aurus ultra, one thing I have noticed recently is the chipset temps were sky high, 87c when playing red dead yesterday after an hour

 

[Metalface Mark]

https://www.3dmark.com/pr/211617

Earlier 3D Mark score today .

 

[Metalface Mark]

As someone that has been benching for the last 20 years or so under both LN2 and Phase, i can tell you this about the pic you posted. It's typical of a cpu that has been overclocked, well overvolted and more importantly not cooled properly to take into count volts and clock.
How do i know this, over the years i've killed more cpu's than most peeps have owned in a lifetime and all of them look like yours. Darkened area (in the case of a 3950x under one CCX), on Intel cpu's it's normally a smaller area but is nearly black as opposed to gold.

Good luck with any RMA and just hope that no one actually checks the cpu.

It has not been overvolted, and overclocked for maybe 4 hours since I got it.

Cooled with a Corsair h115i

under auto volts I’ve seen as low as 1v and as high as 1.52v, I’ve never ran higher than 1.35v manually but the temps were in the high 80s , which is why I put it back to stock

 

[Metalface Mark]

jesus lol, and I thought the pch on my board doing 70c was hot

I bought a pci-e adaptor for one of my m.2 drives (I have 3 in the board) I thought having 3 might be the problem with the pch temps, which led to me doing a stress test before it died, dropped a couple of c but was still high, gigabyte reckoned it’s fine past 100c though.

 

[Metalface Mark]

Can we be sure it's either the cpu or board, I assume you have fully excluded the gpu and ram? Almost sounds more GPU to me giving what you were doing when it went.

will try my gpu in my other pc tomorrow, could be I guess since the fan is running constant 100% when I switch on.

got a new cpu coming tomorrow to test tho, would be a bit of a ball ache to test the gpu since my other pc is all custom watercooled

 

[Metalface Mark]

Can you get into the bios? Try looking at the information centre panel

no, it won’t post at all

 

[Metalface Mark]

Will find out tomorrow when the new cpu comes, not had any issues at all with it prior to this, gpu is overclocked so I guess that’s possibly a suspect too, it’s only really the colour of those pills that’s making me think it’s cpu

 

[Metalface Mark]

welp, replacement cpu no different, tried without the gpu and nothing so it must be the board.

 

[Metalface Mark]

Must be? Could still be the gpu or ram no? When you say tested without the gpu do you mean with a different gpu? Its not going to post without a gpu?

The motherboard has onboard HDMI on it

 

[Metalface Mark]

ah gotcha, will stick an old card in to check.

 

[Metalface Mark]

So took the whole pc apart and trying one component at a time.

Took the cpu out and cleaned it all, then took the motherboard out and dusted it thoroughly to make sure there was nothing shorting anywhere, maybe a loose bit of thermal paste or something, there was a couple of tiny spots around where the cpu went.

Just need tidy everything up again, cables everywhere, but happy to be up and running again without having to RMA anything.

 

[Metalface Mark]

Yeah, back working a treat with the original 3950 in it, very weird!

 

[Metalface Mark]

Temps in the high 80's for 4 hours ? 1.35v with what is plainly not good enough cooling is enough to do that damage to a 7nm cpu.
The voltages you see under "Auto" are not sustained voltages. To give you an example, just running PBO on mine with a 360 AIO cooler i get an all core clock in CB20 of 4.18Ghz and a sustained vcore of 1.215v to 1.219v and temps of 65c to 68c.
I can also get an all core clock of 4.4Ghz at 1.35v, but temps go into the high 70's, not the high 80's. That is where your problem is, too high a temperature, as proved by the burning on the back of the cpu.


What motherboard is it ? I'm shocked that someone that has spent probably getting on for £1300+ on a GPU and £750 on a CPU didn't know that his cpu didn't have an onboard GPU

Im not an idiot, i know the voltages are not sustained. Temperatures are not too high, 3950's are designed to work in the 80's.

CPU has never been near the thermal throttling limit of the CPU. Sorry to disappoint that the CPU wasnt broken, i know you had been rooting for it.

And no, shocking as it seems, as i never planned on ever using onboard graphics, i never checked whether the 3950 was able to do this, so it really doesnt say anything about me not knowing that.

 

[Metalface Mark]

Appreciate the help guys

 

[Metalface Mark]

it was conductonaut i have on the chip.

 

[Metalface Mark]

All i can think is when i switched out to my new case about a week ago that i got something on there, after i took everything out, i brushed everything down, cpu and motherboard, tried it again, no dice, then just before i was about to pack up the board to go back i wiped all around the socket and even in the socket with a damp cloth, cleaned it off then blew dry it with a hairdryer - it came to life after that

 

[Metalface Mark]

So basically your 3950x did not die, you just accidentally got thermal paste where it shouldn't go.

Where is that other dude who claimed your 3950x died because you overclocked it, not a single word now huh

Gave me a chuckle