News Users Report Ryzen 7000X3D Chips Burning Out, Killing Motherboards

[TerryLaze]

All companies have blind spots in their QA process. With so much stuff controlled by firmware and software, catastrophic failure is always one bad update away regardless of who you buy stuff from. AMD may have just had its first major such failure, Intel could be next.

Intel having consistently fewer weird quirks is likely due to having a more extensive regression test suite to run against its CPUs and chipsets.

The x3d chips lack security features that prevent them from burning up, that's a conscious decision to leave them out (or the only way they could do it) .
It's not QA or an oversight.

"...Due to the elongated shape of LGA 1700 CPUs, and how they are secured in the socket, the CPUs have been known to bend or warp while installed in the motherboard’s socket. This can cause higher temperatures due to uneven contact with CPU coolers.

There are a few solutions available for enthusiasts who wish to fix this issue in pursuit of lower CPU temperatures, but they all involve potentially warranty breaking modifications. ..."

https://www.tomshardware.com/reviews/thermalright-lga1700-bcf-contact-frame

Yeah, marvelous, encountering a cooling issue out of the box, and then being faced with void warranty when one wants to fix that...

Yes, being a few degrees hotter than what users would like is the same as the CPU blowing up....

 

[Amdlova]

I build computers from 1994 to this date. Every amd computer come back after some months. Intel hardware have come back to change to a new one. A cheap board tends to last five to eight years a good board last longer.
I have one RMA for the ryzen 5700G (stucking on windows freezing. CPU has seven months no overclock).
My 3600x (given to a friend) and 2700x have random blue screen of death. Tried change every configuration on power or bios.
My computer for now is a old asus x99 with a used xeon has working 70.080 hours and still kicking.

 

[watzupken]

I've never owned an AMD build and most likely never will. AM4 had more than its fair share of bugs and AM5 is no different. It's team blue and team green for me. I'm all about plug & play.

When the power connector on the Nvidia card is melting, it is not a problem. It is a user problem. When AMD GPU and CPU have some problem, it is always AMD driver and stability. In this case, I don’t think there is any conclusion as to whether it is a motherboard issue, but it seems like you have already concluded.

 

[Why_Me]

When the power connector on the Nvidia card is melting, it is not a problem. It is a user problem. When AMD GPU and CPU have some problem, it is always AMD driver and stability. In this case, I don’t think there is any conclusion as to whether it is a motherboard issue, but it seems like you have already concluded.

Remember when it was found to be user error for those connectors melting. It's called being capable of properly plugging in a gpu while not putting those cards in cases that were too small for that gpu to begin with. On the plus side it's always humorous when peeps that have SFF cases purchase that card before checking the specs.

 

[TerryLaze]

Remember when it was found to be user error for those connectors melting. It's called being capable of properly plugging in a gpu while not putting those cards in cases that were too small for that gpu to begin with. On the plus side it's always humorous when peeps that have SFF cases purchase that card before checking the specs.

While I agree with you it's still bad design from nvidia side, it shouldn't be possible to plug in a plug halfway and have the card work, that's why for many years now there are special connectors that have some shorter pins so a device has to be fully plugged in to get power.

I don’t think there is any conclusion as to whether it is a motherboard issue, but it seems like you have already concluded.

Even if it is a motherboard issue it's still a CPU issue because the x3d CPUs have no protections even though every other CPU from AMD does.

 

[outsider2k21]

I build computers from 1994 to this date. Every amd computer come back after some months. Intel hardware have come back to change to a new one. A cheap board tends to last five to eight years a good board last longer.
I have one RMA for the ryzen 5700G (stucking on windows freezing. CPU has seven months no overclock).
My 3600x (given to a friend) and 2700x have random blue screen of death. Tried change every configuration on power or bios.
My computer for now is a old asus x99 with a used xeon has working 70.080 hours and still kicking.

Serious.. X99 is king. Before I upgrade to 7950X last Dec, I was using i7 5820k than 6900k (Found Used 6900k for 150 dollars and used it for 10 months before i upgrade to 7950x) and Asus X99-pro. Honestly can't tell the difference unless I am running benchmarks.

Now it is difficult to find a motherboard come with 10 sata port and consumer cpu that has 40 pcie lane.

 

[sitehostplus]

Mine died this morning.

https://forums.tomshardware.com/threads/new-am5-build-just-died….3804877/

Was playing card games on Steam on default non-OC settings. Will follow up in that thread.

Wow, that stinks.

Hopefully, you bought the extended warranty insurance (Newegg offers Allstate insurance, and I bought it) so it's just a pain in the rear, and not the wallet!

 

[Deleted member 2838871]

Wow, that stinks.

Hopefully, you bought the extended warranty insurance (Newegg offers Allstate insurance, and I bought it) so it's just a pain in the rear, and not the wallet!

I’ve already done an RMA for defective and it said the cost would be refunded?

 

[The Historical Fidelity]

If you do change your mind this board is a steal at that price which includes PCIe 5.0 along with Realtek 4080 audio codec. $590 for the pair.
https://pcpartpicker.com/product/8m6p99/msi-pro-z790-a-wifi-atx-lga1700-motherboard-pro-z790-a-wifi

They just put this cpu on sale.

https://pcpartpicker.com/product/8VqPxr/intel-core-i7-13700-21-ghz-16-core-processor-bx8071513700

Intel???…GROSS!!!

Just kidding haha

 

[The Historical Fidelity]

I build computers from 1994 to this date. Every amd computer come back after some months. Intel hardware have come back to change to a new one. A cheap board tends to last five to eight years a good board last longer.
I have one RMA for the ryzen 5700G (stucking on windows freezing. CPU has seven months no overclock).
My 3600x (given to a friend) and 2700x have random blue screen of death. Tried change every configuration on power or bios.
My computer for now is a old asus x99 with a used xeon has working 70.080 hours and still kicking.

I currently have a 5950x, gave my brother my old 5900x, built my mom a 5700g PC, and built my friend a pc using my older 2700x. All of these are still performing excellently and without trouble. You sure you don’t just suck at building computers?

 

[The Historical Fidelity]

Serious.. X99 is king. Before I upgrade to 7950X last Dec, I was using i7 5820k than 6900k (Found Used 6900k for 150 dollars and used it for 10 months before i upgrade to 7950x) and Asus X99-pro. Honestly can't tell the difference unless I am running benchmarks.

Now it is difficult to find a motherboard come with 10 sata port and consumer cpu that has 40 pcie lane.

Nah fam, X79 is where it’s at. Bought an Asus sabertooth x79 mobo and a used Sandy Bridge Extreme 3960x (got it for 10% of msrp IE paid $100) and have had it all core overclocked by 52% (IE from 3.3 to 5Ghz) for 7 years now and it’s still going strong today.

 

[InvalidError]

Nah fam, X79 is where it’s at. Bought an Asus sabertooth x79 mobo and a used Sandy Bridge Extreme 3960x (got it for 10% of msrp IE paid $100) and have had it all core overclocked by 52% (IE from 3.3 to 5Ghz) for 7 years now and it’s still going strong today.

Today though, you could upgrade to an i3, get roughly twice the performance while using less than half as much power and recover the upgrade cost through reduced power bills over the next couple of years.

 

[The Historical Fidelity]

Today though, you could upgrade to an i3, get roughly twice the performance while using less than half as much power and recover the upgrade cost through reduced power bills over the next couple of years.

True, but sandy bridge will always be held close to my heart because it was the last great overclocking series before Intel switched to the finicky 3D fin transistors. 50% + overclocking will never be a thing again.

 

[klavs]

I haven't checked the details yet, but I suspect that the motherboards (of the users who experience burnout), don't supply the CPU (or the person overclocking) with the correct information with regards to voltage/powerdraw.

Last time this was an issue, approximately two and half years ago, with the Ryzen 3000 line, it turned out that some motherboards underreported the power it delivered to the CPU. Maybe the manufacturers did it to win overclocking tests. Who knows. A few delivered less power - maybe to win efficiency tests? Who knows. Maybe they were just bad quality or due to incompetence.

As far as I recall MSI was the closest to delivering the power that it was supposed to, while ASROG and ASUS were the worst.

 

[Deleted member 14196]

This is just another reason I don’t bother to game on the PC. Too unpredictable with all the wide variation in hardware to consider. My consoles don’t overheat and blow up. That is a really nice feature.

 

[Deleted member 2838871]

D@mn. That was a new build wasn't it?

Results are in…

old mobo and cpu = dead

System powers on fine with new mobo and no cpu and remains dead with old mobo and no cpu.

That being said… Anyone have any thoughts on which CPU I should go with? Stay with 7950x3D? Go with 7950x?

70/30 gaming productivity which is mainly video/editing and encoding of 6k drone footage and blu ray discs.

 

[-Fran-]

Talk about AMD screwing up things with motherboard makers...

I'm surprised AMD hasn't built into AGESA more locks for future CPU releases and treat them as completely different, unsupported, CPUs so the motherboard wouldn't even boot up.

This should have been preventable if motherboard makers would just be a tad more careful with their CPU settings and AMD having better oversight over the offerings.

I say this after watching Hardware Unboxed and the shenanigans they had to go through with the AsRock A620 motherboard. I'm fully convinced this is motherboard makers either not following up AMD's guidance or AMD just being bad at communicating/interacting with them.

TL;DR: AMD is not free from fault here, but hardly the only guilty party. I hope affected people get their replacements hassle free.

Regards.

 

[Pete Mitchell]

The more important question is how are Russian YouTubers getting their hands on AMD CPUs and motherboards? Need to shut that <Mod Edit> down.

 

[TerryLaze]

Last time this was an issue, approximately two and half years ago, with the Ryzen 3000 line, it turned out that some motherboards underreported the power it delivered to the CPU. Maybe the manufacturers did it to win overclocking tests. Who knows. A few delivered less power - maybe to win efficiency tests? Who knows. Maybe they were just bad quality or due to incompetence.

Eh, we all know the truth here, if you don't have to be 100% accurate because you are building something for the military or for science or something like that then you won't be 100% accurate.
This is probably down to them using whatever components they can get their hands on in a big enough number, if it is a little bit too much or a little bit too low won't matter because a little bit of deviance won't blow your CPU up... oops.

I'm surprised AMD hasn't built into AGESA more locks for future CPU releases and treat them as completely different, unsupported, CPUs so the motherboard wouldn't even boot up.

Because these are probably the exact same CPU just with additional cache and not a different batch of CPUs so they probably can't change the cpuID, unless that's nonsense I'm just guessing here.

 

[InvalidError]

Because these are probably the exact same CPU just with additional cache and not a different batch of CPUs so they probably can't change the cpuID, unless that's nonsense I'm just guessing here.

Ever heard of fuse bits and PROMs? You fab a chip with all of the fuse bits intact, then blow up whichever are needed to turn off defective parts of a chip, set the ID to whatever corresponds to the product bin it ends up in, then blow up the master fuse bit to lock everything down permanently.

 

[Amdlova]

Today though, you could upgrade to an i3, get roughly twice the performance while using less than half as much power and recover the upgrade cost through reduced power bills over the next couple of years.

16 cores 8 dimm 3 fans two ssd a gtx 1650 with 3 monitors the idle 85w (gpu eating 33w) full load 225w. Need a 12700 to do same job

 

[TerryLaze]

Ever heard of fuse bits and PROMs? You fab a chip with all of the fuse bits intact, then blow up whichever are needed to turn off defective parts of a chip, set the ID to whatever corresponds to the product bin it ends up in, then blow up the master fuse bit to lock everything down permanently.

That's what I'm saying, if they use perfectly fine working non-3xd cache chips and turn them into 3xd cache chips then wouldn't the cpuID already be "set in stone" ?

 

[CelicaGT]

Lots of conjecture and anecdotes in here. I know it's fun to guess but that is all any of us can do at this point. Not a single AMD, MSI or ASUS engineer in the room and I'm pretty sure none of you are trained in post failure analysis at this level (I am, but for mechanical systems only) so maybe wait for the actual experts to weigh in on the issue before erupting into a fanboi war...

 

[UWguy]

All companies have blind spots in their QA process. With so much stuff controlled by firmware and software, catastrophic failure is always one bad update away regardless of who you buy stuff from. AMD may have just had its first major such failure, Intel could be next.

Intel having consistently fewer weird quirks is likely due to having a more extensive regression test suite to run against its CPUs and chipsets.

True but AMD already has a poor track record as of late with their overheating 7900XTX.

Let's hope they aren't getting drunk on profits and their over inflated stock price and cutting corners (think Boeing).

 

[UWguy]

Results are in…

old mobo and cpu = dead

System powers on fine with new mobo and no cpu and remains dead with old mobo and no cpu.

That being said… Anyone have any thoughts on which CPU I should go with? Stay with 7950x3D? Go with 7950x?

70/30 gaming productivity which is mainly video/editing and encoding of 6k drone footage and blu ray discs.

More evidence it’s likely to be a CPU defect.

Here’s to hoping you get it fixed soon.