News Two Ryzen 7 9800X3D CPUs burned out on X870 motherboards — vendor investigates the Ryzen burnout issues

[Elusive Ruse]

It looks like user error, just my opinion. I haven't built an AM5 system so I do not know what it is like to socket an AM5 cpu, but if it were a major issue I think there would be more reports of it as that MB is one of the best selling boards. Maybe someone who has built an AM5 system can comment on how easy it would be to install the cpu incorrectly. Something is a miss, If that cpu socket was like that out of the box I would have definitely noticed it.

I found socketing AM5 easier and safer since the pins are no longer on the CPU, all I had to do was align the CPU over the socket and let go and it just fell in place. I had to remove my CPU again recently to clean it throughly as I switched to using a Kryosheet instead of thermal paste. Needless to say the second time it was even easier because I knew what to expect.
Installing my Zen had me on my toes though, even after a few times of doing it I was still had to go through my own checklist in my head.

 

[Elusive Ruse]

The 12vhpwr doesn't have a latch?

8s have it and they are easy to insert and ensure they are locked in place. IIRC from the investigation videos done at the time, the first batch of 12s had shabby latches which either needed excessive force to lock in or didn’t make that audible click sound.

 

[redgarl]

100% you are true , he insert cpu in wrong direct , then push cpu down hard (cpu,socket bend)

I dind't catch it at first, but you are correct. He inversely inserted the CPU and locked it in which made the CPU misaligned and crushed the plastic frame of the socket.

 

[redgarl]

AMD users will do anything but blame the obvious culprit for the products failure.

Remember when their 7800x3D exploded, they still blamed the mobo manufacturer.

The common denominator in both cases has been AMD.

It was actually Asus and Gigabyte boosting voltages to get the motherboard crown for performance that was the issue. Funny enough, after AMD pushed new BIOS for AIBs to not been able to do that with x3D, there never was a single more occurance.

Not to mention, all these issues were on ASUS and Gigabyte motherboards only.

So stop speculating and go back to your 14th gen fiasco.

 

[COLGeek]

Please bring the temperature down a bit, folks. No reason to throw stones at each other. Cool?

 

[Auxityne]

"Could be user error!"

Given that one photo shows a bent retention plate, yeah! I'd say so!

 

[Makaveli]

AMD users will do anything but blame the obvious culprit for the products failure.

Remember when their 7800x3D exploded, they still blamed the mobo manufacturer.

The common denominator in both cases has been AMD.

lol came here to troll with your first post.

Well done.

The cause of this was already investigated it was 100% motherboards with increased voltage.

 

[hotaru251]

LGA sockets are also used by Intel, and have been known for being fragile during installation since forever, but once it’s in they’re generally fine and never need touched. PGA sockets cannot achieve the same pin density, so the only other option at the moment is soldered CPUs.

12VHPWR was a replacement to an existing standard that was considered very reliable and was already capable of delivering 600W, albeit with a very bulky 4 connectors. It has a low safety margin, provides less feedback than the previous standard, things like “managing your cables after connecting them” can cause the connector to back out enough for catastrophic failure, and there’s been multiple recalls for bad cables and adapters.

also there is a HUGE difference in that 1 will just kill a cpu while the other can literally have potential to burn your house down as it can literally catch on fire.

That's a very common misconception but it's not true. Once it's plugged in properly, you cannot accidentally "back it out". I don't know where people get this information from but it's not true.

except it can if the cable is messed with afterwards.
You can find videos about it from random people to proper testing like gamersnexus where they show that if you plug it in fully and it clips there is enough "slack" in that clip for some that pushing a cable afterwards can pull it out. (and many times people do this because people don't like rats nest of cables and like it to look nice)

8pin's clip was FIRMLY clipped. you could bend cable w/ force as close to the plug as you wanted and that was never going to come out. (more 8pin gpu's needed made it even less likely due to rest having tension from the close proximity of the other)

12vhp's (at least version 1) specs were unrealistic in majority of cases. (as they are not to be bent from a decent length from connector) and again the slack in clip is not ideal. (likely doesnt help that 12vhp had more than 1 design where as 8pin were basically same design wise)
Most cases (Full towers mostly what actually have enough room for spec) your GPU is too close to the side panel to follow that spec.

 

[tommo1982]

Some have noted the appearance of some pins being bent, which again suggests that poor electrical contact could've been the culprit. The CPU doesn't appear to have any deformation that would've caused the pin bending, so they were probably bent to begin with.

I've seen pics of the socket from Reddit user and there is a sign of damage to it from improperly seated CPU. Another user noted, the top cover is bent near the hinges. It's clearly an user error.
Additionally, the damage is more significant than with the previous X3D. With how much current passes through the CPU's these days, he should be happy only his CPU and mainboard is damaged and nothing cought on fire.

 

[stuff and nonesense]

Nvidia 12vhpower not installed correctly - nvidia should have predicted that people will fail to connect it and should have made it safer.

Amd's socket killing chips, oh well it's the user not installing it correctly.

Okay.

It’s keyed, in a similar way to Intel chips being keyed. Zero insertion force means zero insertion force. If you have to push you are doing it wrong.

Also… RTM

 

[YSCCC]

Guy is pushing 1.2k watts over it, spins the whole PSU around dragging the cable around with him, stabs it with a knife, does all kind of unbelievable things to it, but cable managing will push it out.

View: https://www.youtube.com/watch?v=7hOVY4nklnY


EG1. I lied, he pushed over 1.5k watts...

View: https://www.youtube.com/watch?v=p0fW5SLFphU&t=362s


go see this, you are finding a straight pull on the cable, which is not any sane person will try to do for any sort of installation or cable management.

De8auer have his cable visually speaking at least to be fully seated, and the bad connection light came off just sslightly pulling it sideways without undo the whole plug, that is what was generally referred to as the pull out, you don't need to unlatch it, but the bending is enough to cause it to have issues.

Now as said before, combined with the gigantic cards and anyone building a PC with reasonable airflow, bending the cable sideways is always required unless you basically run an open bench. combined with the fact that the click isn't as easily noticeable it is a design flaw if not overlook. Also that's why the new PSU specs recommend to say the least, don't bend it before 35mm clearance (which doesn't have in most PC case).

Meanwhile when compared to putting the CPU in the wrong position without even drop it in and try wiggle it a bit to find it fully seated before clamping is more akin to ppl just brute force the ram in opposite direction and snap the pins (which, do happens once a while), is the issue where ppl complain about the Nvidia Ada cards at the launch, especially back then the 4090 basically cost more than all your other PC parts combined. And don't forget, among the first reoirted melting connector, ppl do also says it's likely user error, then when the issue worsen with time, more investigations have revealed all the flaws and resulted in the 12v 2x6 version we have right now. To claim to be neutral as you did, do consider not bashing for a new single case and hope the GN study likely in a few weeks time to see if we have more things than improper install and clamping of the CPU in an LGA socket (where Intel did have fried CPUs back in the days, but none really keep bashing LGA is a bad idea).

If there are more 9800X3D frying itself and not the normal Zen 4 or 5, it should be some kind of hidden flaws, but if it isn't? this is just an example of brute forcing installation without reading the manual

 

[snapdragon-x]

One more case in South Korea: https://quasarzone.com/bbs/qn_hardware/views/1751432
Same board and CPU.

 

[jlake3]

Ah, so you agree that he was wrong and that the 12vhpwr don't come out. Shall we ask them for evidence or it's fine?

Partially backed out, burned connectors have been reported by several sources, probably most notably Gamer’s Nexus.

View: https://www.youtube.com/watch?v=Y36LMS5y34A&t=1621s


JayzTwoCents I know has also covered roundups of posts where someone seated it, did their cable management, then had a slight gap in the connection.

PCI-SIG actually changed the spec to shorten the sense pins, as a countermeasure against systems attempting to pull full draw over a loose connector. Multiple vendors are making brightly colored connectors so users can check their seating. Vendors are still advising you do not bend the cable within 35mm of the connector to avoid placing stress the pulls the pins out and causes the contact area to drop.

I have two PSUs with the new connector, bought in anticipation of an upgrade that hasn’t happened yet. My experience is that the clips are stiffer and less tactile, which matches the experience others have reported.

You’ve clearly made up your mind on that 12VHPWR good, AM5 bad, though.

 

[YSCCC]

One more case in South Korea: https://quasarzone.com/bbs/qn_hardware/views/1751432
Same board and CPU.

Yea, saw that also, and funny part is both have a basically the same pattern yet opposite direction damage on the CPU socket plastic part, buildzoid have quite an interesting (albiet not really friendly tone) view on how is it possible to get it damaged this way.

View: https://www.youtube.com/watch?v=qY0kEB-1MIc

 

[tommo1982]

One more case in South Korea: https://quasarzone.com/bbs/qn_hardware/views/1751432
Same board and CPU.

Check the hinges, they are bent. The pressure plate was damaged during installation.

 

[epiczombiekill]

Yet another reason to never buy MSI
Also if you have a CPU burn out you should never put a new one in the same motherboard because it's most likely the MB was the culprit, if the CPU didn't just take it out

 

[Avro Arrow]

I saw a Jayz2Cents video about this. He believes that it's user error for the same reasons as Aaron. What was funny though was a different take on the ID10.T error. He called it "P.I.C.N.I.C." which stands for "Problem in chair, not in computer". I had never heard that one before so I admit it, I LOLed.

 

[stuff and nonesense]

"Problem in chair, not in computer".

Seat to keyboard interface..

 

[Eximo]

I saw a Jayz2Cents video about this. He believes that it's user error for the same reasons as Aaron. What was funny though was a different take on the ID10.T error. He called it "P.I.C.N.I.C." which stands for "Problem in chair, not in computer". I had never heard that one before so I admit it, I LOLed.

Also Pebcak. Problem Exists Between Chair and Keyboard.

 

[Eximo]

I want to say in a very recent Jayz2Cents video they accidentally forgot to plug a 12VHPWR connector in all the way while they were testing an Arrowlake system?

 

[redgarl]

Yet another reason to never buy MSI
Also if you have a CPU burn out you should never put a new one in the same motherboard because it's most likely the MB was the culprit, if the CPU didn't just take it out

You are a genius or something?

It is not MSI's fault, it is the fault of the user who installed the CPU backward without noticing it.

View: https://youtu.be/qY0kEB-1MIc

 

[snapdragon-x]

Check the hinges, they are bent. The pressure plate was damaged during installation.

Yeah, I think that's most likely the best explanation. I thought it was pretty hard to get that wrong if you're not a preschooler anymore...

 

[snapdragon-x]

I want to say in a very recent Jayz2Cents video they accidentally forgot to plug a 12VHPWR connector in all the way while they were testing an Arrowlake system?

I scrubbed quickly through AL videos but didn't see it. Which video was that?

 

[YSCCC]

Check the hinges, they are bent. The pressure plate was damaged during installation.

I wonder how can one clamp the ILM to bend even that part, or even dare to use the ILM in that shape if it's shipped like that

 

[newtechldtech]

How do you know the user didn't initially put it in wrong, then fix it before installing the heatsink and turning it on? That could've been enough to cause the damage leading to this outcome.

The "burn marks"