The Truth About AMD's CPU Failures: X-Ray, Electron Microscope, & Ryzen Burns
video description
Date: 2023-05-10
Related videos
Comments and reviews: 15
Cyber
Thanks for this detailed analysis from the GN team, I really appreciate and value this type of research. I've only worked with PCB failure analysis before to a limited extent as a PCB designer that previously functioned as a systems engineer.
Personally I'd also like to perhaps contribute some input for consideration as I didn't quite see some of the simpler failure mechanics elaborated upon. The PCB interposer deformation and delamination near the over-current event site is quite typical of extreme localized heating for PCB's. Copper when significantly heated in a localized area thermally expands, while the outer colder areas constrained it, this causes buckling and subsequent delamination of the area. Meaning the outer, least compressed copper layers warp away from the PCB core and deform the socket pins. The socket pins due to extreme heating also loose their elasticity springyness as this effectively causes the crystal structure of the metal to change and accept the new depressed shape as normal once cooled.
Other things that occur inside the PCB layers as a result of extreme localized heating is off-gasing of more volatile compounds that have a low phase transition temperature in addition to vaporization of moisture into steam inside the PCB. There is always just a tiny bit trapped from earths atmosphere, PCB materials are a little bit permeable to an extent. This all combines to create both mechanical buckling of the outer layers and possible vapour expansion voids within the PCB which we in the Industry like to call Popcorning due to it's distinctive sound . This often breaks multiple current carrying paths in the area or forms new ones due to shorts which often results in a runaway failure as current is conducted over ever fewer points thus creating more failures.
As for the dies exhibiting voids in the indium-alloy solder layer, I'm not really qualified in that domain, but I suspect similar processes may occur here due to vapour formation either during soldering as part of the flux used which is not escaping effectively during the short soldering process. As well as potentially new voids (gas bubbles) introduced when the silicon overheated to such an extent as to crack and reflow the thermal interface solder. During this process materials from below the silicon die may be wicked up through cracks to the top.
I have in on instance ended up with underfill material on-top of a chip I was destructively testing.
Hope this can add what little extra context I can provide.
reply
Thanks for this detailed analysis from the GN team, I really appreciate and value this type of research. I've only worked with PCB failure analysis before to a limited extent as a PCB designer that previously functioned as a systems engineer.
Personally I'd also like to perhaps contribute some input for consideration as I didn't quite see some of the simpler failure mechanics elaborated upon. The PCB interposer deformation and delamination near the over-current event site is quite typical of extreme localized heating for PCB's. Copper when significantly heated in a localized area thermally expands, while the outer colder areas constrained it, this causes buckling and subsequent delamination of the area. Meaning the outer, least compressed copper layers warp away from the PCB core and deform the socket pins. The socket pins due to extreme heating also loose their elasticity springyness as this effectively causes the crystal structure of the metal to change and accept the new depressed shape as normal once cooled.
Other things that occur inside the PCB layers as a result of extreme localized heating is off-gasing of more volatile compounds that have a low phase transition temperature in addition to vaporization of moisture into steam inside the PCB. There is always just a tiny bit trapped from earths atmosphere, PCB materials are a little bit permeable to an extent. This all combines to create both mechanical buckling of the outer layers and possible vapour expansion voids within the PCB which we in the Industry like to call Popcorning due to it's distinctive sound . This often breaks multiple current carrying paths in the area or forms new ones due to shorts which often results in a runaway failure as current is conducted over ever fewer points thus creating more failures.
As for the dies exhibiting voids in the indium-alloy solder layer, I'm not really qualified in that domain, but I suspect similar processes may occur here due to vapour formation either during soldering as part of the flux used which is not escaping effectively during the short soldering process. As well as potentially new voids (gas bubbles) introduced when the silicon overheated to such an extent as to crack and reflow the thermal interface solder. During this process materials from below the silicon die may be wicked up through cracks to the top.
I have in on instance ended up with underfill material on-top of a chip I was destructively testing.
Hope this can add what little extra context I can provide.
reply
nighthawk475
Just to be clear, since it wasn't reiterated in this new upload (but was in part 1) - the best understanding we have today is still that things are safe now for all 7000 series processors so long as you're on an updated/latest bios for am5 motherboards? (Assuming you're buying from a standard manufacturer).
I've seen some people posting pictures of a couple lightly burned ryzen processors lately too, where there was visible discoloration on the underside, but no visible melting/shattering like this original example. It'd maybe be reassuring to see you try to up the SOC too high on a more recent bios to see if the failsafes are any better. (and if ASUS has more issues I'd be very curious to see how other brands compare, since they haven't been getting scrutinized as much after all this, I feel)
reply
Just to be clear, since it wasn't reiterated in this new upload (but was in part 1) - the best understanding we have today is still that things are safe now for all 7000 series processors so long as you're on an updated/latest bios for am5 motherboards? (Assuming you're buying from a standard manufacturer).
I've seen some people posting pictures of a couple lightly burned ryzen processors lately too, where there was visible discoloration on the underside, but no visible melting/shattering like this original example. It'd maybe be reassuring to see you try to up the SOC too high on a more recent bios to see if the failsafes are any better. (and if ASUS has more issues I'd be very curious to see how other brands compare, since they haven't been getting scrutinized as much after all this, I feel)
reply
dragonh
Gamers Nexus
I had an issue when I was building my new 7950x3d, I went with asus first time, it worked fine for the first 5 days, but then windows wouldn't boot anymore, then after a few more tries, the BIOS wouldn't boot anymore. It kept giving me a CPU light.
Eventually after playing around with it a lot, I couldn't get it to boot still, so I went with gigabyte and that worked fine and still does.
After the removed the CPU from the asus motherboard, I did see some sort of damage in the pins, it almost looked like a burn? And I saw this shiny sheen across my 7950x3d.
Luckily the CPU still worked, it's been about a month since that event.
You guys want my sample? I still have the mobo just sitting there, though I am using the CPU.
reply
Gamers Nexus
I had an issue when I was building my new 7950x3d, I went with asus first time, it worked fine for the first 5 days, but then windows wouldn't boot anymore, then after a few more tries, the BIOS wouldn't boot anymore. It kept giving me a CPU light.
Eventually after playing around with it a lot, I couldn't get it to boot still, so I went with gigabyte and that worked fine and still does.
After the removed the CPU from the asus motherboard, I did see some sort of damage in the pins, it almost looked like a burn? And I saw this shiny sheen across my 7950x3d.
Luckily the CPU still worked, it's been about a month since that event.
You guys want my sample? I still have the mobo just sitting there, though I am using the CPU.
reply
Ben
I was excited when Steve said 'This is one of the coolest videos we ever made' but damn that is somehow selling it short! Seriously one of the best pieces of content I've ever seen period. I'm so glad to hear that Steve and the entire GN team enjoyed putting in all the work to get us to this point and hope that we may be graced with similar analysis in the future! Adding this on top of Steve's ability to hold companies PR feet to the fire when they want to sweep something under the rug; I look forward to him dissecting AMD/ASUS/MB'S communications regarding the AM5 failures that he had more knowledge of than anyone outside of those companies.
reply
I was excited when Steve said 'This is one of the coolest videos we ever made' but damn that is somehow selling it short! Seriously one of the best pieces of content I've ever seen period. I'm so glad to hear that Steve and the entire GN team enjoyed putting in all the work to get us to this point and hope that we may be graced with similar analysis in the future! Adding this on top of Steve's ability to hold companies PR feet to the fire when they want to sweep something under the rug; I look forward to him dissecting AMD/ASUS/MB'S communications regarding the AM5 failures that he had more knowledge of than anyone outside of those companies.
reply
Fender178
Steve I have a few questions here. Who do you think is responsible for the catastrophic CPU failure here? Is it AMD? The motherboard manufacture? Because in the 1st part you speculated both parties could be at fault here. Do you still hold that opinion after doing this video? Otherwise very good job on this video series so far. Can't wait to see what you say about ASUS and if you get put on their crap list like you did with MSI. I like the how much detail you go into with all of your in-depth videos covering things from the Gigabyte exploding PSUs to the melting 12V High powered cables to your coverage on Artisan builds.
reply
Steve I have a few questions here. Who do you think is responsible for the catastrophic CPU failure here? Is it AMD? The motherboard manufacture? Because in the 1st part you speculated both parties could be at fault here. Do you still hold that opinion after doing this video? Otherwise very good job on this video series so far. Can't wait to see what you say about ASUS and if you get put on their crap list like you did with MSI. I like the how much detail you go into with all of your in-depth videos covering things from the Gigabyte exploding PSUs to the melting 12V High powered cables to your coverage on Artisan builds.
reply
Pac
i would not recommend Warframe to my worst enemy. its a great game, i got to a point where i was soloing tridolons and helping new players farm for new stuff but the chats mods are toxic and they bully people for their own amusement. I've been banned and unbanned twice without a reason and without an apology. I've witnessed R3D PO1NT, Server, Chimp, Lazotto all bully people for asking basic questions in the Q&A tab. The devs do nothing and blame you for the issue and suggest you dont bother their moderators
reply
i would not recommend Warframe to my worst enemy. its a great game, i got to a point where i was soloing tridolons and helping new players farm for new stuff but the chats mods are toxic and they bully people for their own amusement. I've been banned and unbanned twice without a reason and without an apology. I've witnessed R3D PO1NT, Server, Chimp, Lazotto all bully people for asking basic questions in the Q&A tab. The devs do nothing and blame you for the issue and suggest you dont bother their moderators
reply
GeFeldz
I wish ABit would have stayed in business and after DFI poached their best engineers, that DFI would have stayed in business... It's great that we essentially have 4 way competition in motherboards (yes, i know there are more brands) for the DIY market, but there sure isn't anything wrong with more players!
In my experience, ASUS has always had good enough boards with a slight premium, however the last few years they've really shown they're just fat and happy and resting on old accomplishments.
reply
I wish ABit would have stayed in business and after DFI poached their best engineers, that DFI would have stayed in business... It's great that we essentially have 4 way competition in motherboards (yes, i know there are more brands) for the DIY market, but there sure isn't anything wrong with more players!
In my experience, ASUS has always had good enough boards with a slight premium, however the last few years they've really shown they're just fat and happy and resting on old accomplishments.
reply
ExplouD
AMD may be partly to blame but ASUS oh my god... they have been making mostly crap since about INTEL's core i 4000 series and not to mention their disastrous software/malware and people still buy from them! I won't deny that they may put out a good product once in a while but overall they are disastrous and people buy their marketing and others have the nostalgic memory when the brand was doing things right when it is no longer like that a little over 10 years ago....
reply
AMD may be partly to blame but ASUS oh my god... they have been making mostly crap since about INTEL's core i 4000 series and not to mention their disastrous software/malware and people still buy from them! I won't deny that they may put out a good product once in a while but overall they are disastrous and people buy their marketing and others have the nostalgic memory when the brand was doing things right when it is no longer like that a little over 10 years ago....
reply
James
Absolutely outstanding work from you guys. There s no hiding things by large companies and the masses have there Sherlock holms in Gamers Nexus to out them bring them to justice and sword fight for the people. Where honour and justice will prevail.
Seriously though guys if could buy the equipment the lab had I would and give to you guys. This is beyond next level of looking out for the people.
Massive thank you is deserved from so many to so few.
reply
Absolutely outstanding work from you guys. There s no hiding things by large companies and the masses have there Sherlock holms in Gamers Nexus to out them bring them to justice and sword fight for the people. Where honour and justice will prevail.
Seriously though guys if could buy the equipment the lab had I would and give to you guys. This is beyond next level of looking out for the people.
Massive thank you is deserved from so many to so few.
reply
The
Very good, detailed analysis and processing. But if we are honest superfluous for the normal customer. The number of affected CPUs remains negligible and the current and upcoming bios updates will fix it for normal home users.
The bios and Asus video will certainly be interesting. Asus has slacked off as a company in many areas and the brand hype hasn't been justified here for a long time. This new example shows it again.
reply
Very good, detailed analysis and processing. But if we are honest superfluous for the normal customer. The number of affected CPUs remains negligible and the current and upcoming bios updates will fix it for normal home users.
The bios and Asus video will certainly be interesting. Asus has slacked off as a company in many areas and the brand hype hasn't been justified here for a long time. This new example shows it again.
reply
Menko
18:55 I've had some limited contact with people who work for silicon IP companies (basically they make features like say auto-overclocking for example, patent it, and sell it to AMD/Intel). I know that some of the people at those companies, stare at verilog-generated silicon for months on end, all day, every day. I'm willing to bet some of those people could identify the part they were working on just from this picture
reply
18:55 I've had some limited contact with people who work for silicon IP companies (basically they make features like say auto-overclocking for example, patent it, and sell it to AMD/Intel). I know that some of the people at those companies, stare at verilog-generated silicon for months on end, all day, every day. I'm willing to bet some of those people could identify the part they were working on just from this picture
reply
Raphael
This is so great, and I think a lot of us are really thankful for the work you're putting in. A question I have though is what does this mean for the end user of a Ryzen 7000 CPU? I run a 7600X and my VSOC is indeed at 1.35, but my mobo manufacturer (MSI) has only released a beta bios so far to correct the issue, which I'm leery of using. What should I do to try and mitigate the issue in the meantime?
reply
This is so great, and I think a lot of us are really thankful for the work you're putting in. A question I have though is what does this mean for the end user of a Ryzen 7000 CPU? I run a 7600X and my VSOC is indeed at 1.35, but my mobo manufacturer (MSI) has only released a beta bios so far to correct the issue, which I'm leery of using. What should I do to try and mitigate the issue in the meantime?
reply
Radio
The Enterprise computer system is controlled by three primary main processing cores, cross-linked with a redundant melacortz ramistat. Fourteen kiloquad interface modules. The core element is based on an FTL nanoprocessor with twenty five bilateral kelilactirals, with twenty of those being slaved into the primary heisenfram terminal. Now you do know what a bilateral kelilactiral is?
reply
The Enterprise computer system is controlled by three primary main processing cores, cross-linked with a redundant melacortz ramistat. Fourteen kiloquad interface modules. The core element is based on an FTL nanoprocessor with twenty five bilateral kelilactirals, with twenty of those being slaved into the primary heisenfram terminal. Now you do know what a bilateral kelilactiral is?
reply
danytoob
I wish you'd give us some details on what happened...oh, wait. Never mind.
Thanks, Steve!
But honestly, hoz'bout a new anniversary coffee mug to match the latest shirt (black mug, blue & gold 15 logo). That would let me start each & every day waking up to dielectric breakdown and substrate tunneling. If that isn't (insert explicative here), I don't know what is. DT
reply
I wish you'd give us some details on what happened...oh, wait. Never mind.
Thanks, Steve!
But honestly, hoz'bout a new anniversary coffee mug to match the latest shirt (black mug, blue & gold 15 logo). That would let me start each & every day waking up to dielectric breakdown and substrate tunneling. If that isn't (insert explicative here), I don't know what is. DT
reply
bob
In all honesty, while it is clearly a problem that this has happened, I can't help but be truly impressed at the astonishing scale engineering and knowledge that has to go into the design and manufacture of chips. And stellar content like this allows me a brief but incredibly meaningful glimpse into that world. Thanks Steve and team. You guys are slaying!
reply
In all honesty, while it is clearly a problem that this has happened, I can't help but be truly impressed at the astonishing scale engineering and knowledge that has to go into the design and manufacture of chips. And stellar content like this allows me a brief but incredibly meaningful glimpse into that world. Thanks Steve and team. You guys are slaying!
reply
Add a review, comment
Other channel videos
