Layers write nuanced statements all the time. What are you even talking about? Most bulges are directly under the CCD, and keyboard experts like you think it’s the VSOC where a little over 1.3V can burn the IO die.
The GamersNexus video backs up what he said about VRMs pushing insane power into a CPU that's in a state where it should be triggering OCP rather than additional power.
Turns out it was companies like Asus not fallowing the guidelines and raising the SoC voltage beyond the safe limits. Apparently, Asus used the Expo settings to do this by adding in their own voltage settings to allow for higher DDR5 OC.
Anyone interested can go to YouTube and watch Gamers Nexus video about this. I guess Asus is also being pretty scummy about it as well towards their customers on this matter. Gamers Nexus has a whole video about it. I am also not sure if there are other board makers that did the same thing as Asus did for the SoC voltage beyond safe limits.
Lmao, chat bots don't reorganize/rewrite statements and then forget to eliminate the duplicate, but it's one of the most common mistakes I do when I'm editing my own articles.
Calling it a "chat bot error" is basically admitting you've never written anything longer than a Twitter post, and yet are trying to judge an actual writer, who's only fault is working for a company that can't afford an editor.
"One interesting point about AMD's statement is that it eludes to whether or not the issue is just on its Ryzen 7000X3D processors or whether it affects all of its Ryzen 7000 processors entirely."
That's unfortunate. CPU competition is good, but both Intel and AMD have been flying a bit close to the proverbial Sun lately and those wax wings are only going to tolerate so much before they melt.
"Gamers and their requirements are why everyone is flying close to the sun."
ah, such a market. it was Lotus 1-2-3 and WordPerfect and DBase III that made the killer app crew. all productivity enhancers. these days it's time waster the PC world is relying on. that's progress, I guess.
well... the PC-gamer market isn't the same thing as crypto-mining (a truly wagering experience) or FanDuel and the like. I suspect that the Big Boys run on IBM Big Iron running linux, not a bunch of PCs running Netware.
Each of those three could have turned me off PCs, if I hadn't been completely smitten by then and running Microport System V release 2 on my 80286, porting code.
It was Visicalc, WordStar and Vulcan aka dBase II for me from the very start on CP/M on my Apple ][ clone with a Z-80 softcard, all those 2nd and 3rd runner-ups just made things worse.
I yielded to M$ office after GEM lost, but tried every alternative from Nico Börries, since he sold the Turbo-Text-Toolbox as a WordStar replacement (that's LibreOffice today, in case you don't know your Personal Computer history).
I still wonder why this issue surfaced after all these months 6 months+ of Ryzen 7000 series AND X3D refresh being a month + already present in the market. There should be more for this. But I still cannot think how can AMD and Mobo partners can skip the SoC voltage specification. This is exactly why I always mention AMD needs to provide datasheets. Intel does it and they have voltage specifications for a lot of variables. Esp given the fact how this is EXPO related or at-least it makes that like it.
So many reviewers and OCers and everyone ran these processors with EXPO. AMD should really need to put focus more perhaps.
It sounds like this is more a case that the X3D is uniquely vulnerable to over-voltage and pver-current conditions but the conditions were still there on the non-vcache chips.
7800X3D here with Asus X670E-A motherboard and Corsair VENGEANCE 32GB (2x16GB) DDR5 DRAM 6000MT/s C36 AMD EXPO Memory Kit SKU CMK32GX5M2D6000Z36. I updated to BIOS 1202 recently but now see "Beta" BIOS 1301 which i'm personally not installing for now. I did turn my memory EXPO II BIOS setting back to Auto until they work out this apparent over CPU Voltage issue with EXPO memory set. I did notice that with EXPO II setting on the memory (I was using the memory BIOS EXPO II (3000 MHz) setting until yesterday when I turned it back to Auto (2400 MHz)) my 7800X3D CPU VDDR_SOC Voltage went from EXPO II Max seen of 1.340 V to Auto Max seen of 1.020 V.
I'm now updated to 1303 BIOS and re-enabled EXPO II settings (DDR5-6000 (3000 MHz) memory). CPU VDDR_SOC MAX is 1.240 V like you had with 1301 BIOS. That seems reasonable. I'll run at this and see how things go.
In addition, I noticed these 7800X3D Package Temps (after gaming a little): EXPO II memory BIOS setting: 72.0 C Max Auto memory BIOS setting: 66.4 C Note: all my data is found using CPUID HWMonitor in Admin mode
What we have here is Mobo suppliers pushing the limits to get to the top of the charts knowing full well the blame would be focussed totally on AMD, its not surprising that the majority of failures seem to be related to (but not confined) ASUS who are seen as a halo brand with ROG and need to maintain that position and justify the premium they charge. AMDs failure was in trusting the Mobo makers to be sensible rather than implementing a hard cap which they have now done, perhaps the refresh later this year can take the voltage without error we will find out later i guess but for now just update the Bios
AMD didn’t learn from previous iterations of motherboard maker shenanigans, like ASRock 970 boards — ostensibly rated for the 9000-series FX chips that caught fire and ASUS’ penchant (via the Crosshair Formula Z) for underreporting the vcore.
Of course, AMD should have been tarred and feathered for the 9000 series in the first place.
For me this is AMD's FDIV-bug moment: they either come out now and promise "life-time" replacement warranties, or there will be far too many current owners wondering if they should now burn up their potentially damaged CPU to get a known-good replacement.
And let's get rid of that silly notion that PBO/EXPO activation kills warranties while they are at it. It's an advertised feature and using that may not affect warranties: you make it safe to use or you leave it out.
I have to say nothing really excites me about this motherboard chipset. They obviously use way to much current/wattage which is what is causing this problem. Maybe instead of simply making things faster they should focus on making them energy efficient. I could say the same for graphic cards. When we approach using over 500 watts of constant power just to feed the beast I say we have gone too far. I wouldn't want the monthly electric bill.
A few days ago, my 7950X unexpectedly failed. I had kept all settings on Auto, except for enabling XMP DDR5-6000. Initially, I experienced random reboots, but they became more frequent over the past week. Eventually, the system failed to boot into Windows, resulting in a blue screen.
I suspected that the issue might be related to my Windows installation or M.2 drive, so I attempted to boot from a USB Windows Installer. However, it would get stuck right after the POST. I tried various USB bootable media, but only MEMTEST86 would launch. Running a test revealed thousands of errors before crashing.
To troubleshoot, I replaced the 7950X with a 7600X from my secondary system and everything worked flawlessly. When I put the 7950X back into the secondary system, it exhibited the same issues, unable to boot into Windows or USB media. Upon inspection, I noticed some darker gold markers beneath the 7950X where the memory controller is located.
In my 30 years of building PCs, this is the first time a CPU has failed on me. As the push for smaller nanometers continues, AMD should exercise greater caution with voltages as the CPUs seem to be a lot more fragile.
We’ve updated our terms. By continuing to use the site and/or by logging into your account, you agree to the Site’s updated Terms of Use and Privacy Policy.
Facebook
Twitter
RSS
42 Comments
Back to Article
jerrylzy - Thursday, April 27, 2023 - link
This statement actually doesn’t say VSOC is the root cause. It’s just one of the power rails getting capped. All indications are VCore related.Gothmoth - Thursday, April 27, 2023 - link
well i do read SOC in the statement and not VCORE.... but the keyboard "experts" here sure know better. :-)Gothmoth - Thursday, April 27, 2023 - link
i guess some like to argue over semantics.jerrylzy - Thursday, April 27, 2023 - link
Layers write nuanced statements all the time. What are you even talking about? Most bulges are directly under the CCD, and keyboard experts like you think it’s the VSOC where a little over 1.3V can burn the IO die.bananaforscale - Friday, April 28, 2023 - link
Says a guy who doesn't understand that an overvoltage can cause a short, and that modern VRMs can easily push hundreds of amps through a short?jerrylzy - Sunday, April 30, 2023 - link
Do you both feel stupid now after watching Gamers Nexus’s root cause video?dqniel - Monday, May 1, 2023 - link
The GamersNexus video backs up what he said about VRMs pushing insane power into a CPU that's in a state where it should be triggering OCP rather than additional power.bananaforscale - Friday, April 28, 2023 - link
VSoC is not Vcore. Have you ever looked at a BIOS and voltage settings?jerrylzy - Sunday, April 30, 2023 - link
I don’t talk to anyone who can’t read.dqniel - Monday, May 1, 2023 - link
No, all indications are that it's vSOC related. Not vCore.rocky12345 - Sunday, May 14, 2023 - link
Turns out it was companies like Asus not fallowing the guidelines and raising the SoC voltage beyond the safe limits. Apparently, Asus used the Expo settings to do this by adding in their own voltage settings to allow for higher DDR5 OC.Anyone interested can go to YouTube and watch Gamers Nexus video about this. I guess Asus is also being pretty scummy about it as well towards their customers on this matter. Gamers Nexus has a whole video about it. I am also not sure if there are other board makers that did the same thing as Asus did for the SoC voltage beyond safe limits.
gryer7421 - Thursday, April 27, 2023 - link
""What this means needs further clarification, as AMD hasn't specified what it means by preventing the CPU from exceeding its specification limits. ""You ghost writting chat bot needs work ...
Gavin Bonshor - Thursday, April 27, 2023 - link
Yeah, that was my fault. I had to get to the doctor quickly as my son wasn't well. I've edited it; thank you for highlighting it.Wereweeb - Thursday, April 27, 2023 - link
Lmao, chat bots don't reorganize/rewrite statements and then forget to eliminate the duplicate, but it's one of the most common mistakes I do when I'm editing my own articles.Calling it a "chat bot error" is basically admitting you've never written anything longer than a Twitter post, and yet are trying to judge an actual writer, who's only fault is working for a company that can't afford an editor.
HarryVoyager - Sunday, April 30, 2023 - link
I remember one time I managed to leave half a sentence from a deleted thought in the middle of a completely different paragraph.The poor editor was so confused.
NaterGator - Thursday, April 27, 2023 - link
"One interesting point about AMD's statement is that it eludes to whether or not the issue is just on its Ryzen 7000X3D processors or whether it affects all of its Ryzen 7000 processors entirely."Gavin, I think you meant alludes?
Ryan Smith - Thursday, April 27, 2023 - link
Yep. Fixed.PeachNCream - Thursday, April 27, 2023 - link
That's unfortunate. CPU competition is good, but both Intel and AMD have been flying a bit close to the proverbial Sun lately and those wax wings are only going to tolerate so much before they melt.Threska - Thursday, April 27, 2023 - link
Gamers and their requirements are why everyone is flying close to the sun. Just look at the current state of GPUs.FunBunny2 - Friday, April 28, 2023 - link
"Gamers and their requirements are why everyone is flying close to the sun."ah, such a market. it was Lotus 1-2-3 and WordPerfect and DBase III that made the killer app crew. all productivity enhancers. these days it's time waster the PC world is relying on. that's progress, I guess.
stephenbrooks - Friday, April 28, 2023 - link
I'm pretty sure the whole GPGPU (using GPUs for science etc.) only exists because of the multi-billion-dollar gaming industry.abufrejoval - Sunday, April 30, 2023 - link
It's a classic synnergy, as science cannot be separated from child's play and vice versa: neither could have done without the other.FunBunny2 - Tuesday, May 2, 2023 - link
" multi-billion-dollar gaming industry."well... the PC-gamer market isn't the same thing as crypto-mining (a truly wagering experience) or FanDuel and the like. I suspect that the Big Boys run on IBM Big Iron running linux, not a bunch of PCs running Netware.
Oxford Guy - Saturday, April 29, 2023 - link
Productivity is more than a waste of time; it’s destroying the biosphere.abufrejoval - Sunday, April 30, 2023 - link
Each of those three could have turned me off PCs, if I hadn't been completely smitten by then and running Microport System V release 2 on my 80286, porting code.It was Visicalc, WordStar and Vulcan aka dBase II for me from the very start on CP/M on my Apple ][ clone with a Z-80 softcard, all those 2nd and 3rd runner-ups just made things worse.
I yielded to M$ office after GEM lost, but tried every alternative from Nico Börries, since he sold the Turbo-Text-Toolbox as a WordStar replacement (that's LibreOffice today, in case you don't know your Personal Computer history).
TheinsanegamerN - Tuesday, May 2, 2023 - link
Those damn gamers and their *checks notes* willingness to buy our products! How DARE they!Silver5urfer - Thursday, April 27, 2023 - link
I still wonder why this issue surfaced after all these months 6 months+ of Ryzen 7000 series AND X3D refresh being a month + already present in the market. There should be more for this. But I still cannot think how can AMD and Mobo partners can skip the SoC voltage specification. This is exactly why I always mention AMD needs to provide datasheets. Intel does it and they have voltage specifications for a lot of variables. Esp given the fact how this is EXPO related or at-least it makes that like it.So many reviewers and OCers and everyone ran these processors with EXPO. AMD should really need to put focus more perhaps.
HarryVoyager - Sunday, April 30, 2023 - link
It sounds like this is more a case that the X3D is uniquely vulnerable to over-voltage and pver-current conditions but the conditions were still there on the non-vcache chips.Jeff72 - Thursday, April 27, 2023 - link
7800X3D here with Asus X670E-A motherboard and Corsair VENGEANCE 32GB (2x16GB) DDR5 DRAM 6000MT/s C36 AMD EXPO Memory Kit SKU CMK32GX5M2D6000Z36. I updated to BIOS 1202 recently but now see "Beta" BIOS 1301 which i'm personally not installing for now. I did turn my memory EXPO II BIOS setting back to Auto until they work out this apparent over CPU Voltage issue with EXPO memory set. I did notice that with EXPO II setting on the memory (I was using the memory BIOS EXPO II (3000 MHz) setting until yesterday when I turned it back to Auto (2400 MHz)) my 7800X3D CPU VDDR_SOC Voltage went from EXPO II Max seen of 1.340 V to Auto Max seen of 1.020 V.krypto1300 - Thursday, April 27, 2023 - link
I updated to 1301 and my SOC voltage dropped from 1.35 to 1.24 with EXPO. Performance is much the same in CB, Timespy and gaming.Jeff72 - Friday, April 28, 2023 - link
1303 BIOS now availableJeff72 - Friday, April 28, 2023 - link
I'm now updated to 1303 BIOS and re-enabled EXPO II settings (DDR5-6000 (3000 MHz) memory).CPU VDDR_SOC MAX is 1.240 V like you had with 1301 BIOS. That seems reasonable.
I'll run at this and see how things go.
Jeff72 - Thursday, April 27, 2023 - link
In addition, I noticed these 7800X3D Package Temps (after gaming a little):EXPO II memory BIOS setting: 72.0 C Max
Auto memory BIOS setting: 66.4 C
Note: all my data is found using CPUID HWMonitor in Admin mode
krypto1300 - Thursday, April 27, 2023 - link
My temps are also lower with auto memory settings becasue voltages are lower.alufan - Friday, April 28, 2023 - link
What we have here is Mobo suppliers pushing the limits to get to the top of the charts knowing full well the blame would be focussed totally on AMD, its not surprising that the majority of failures seem to be related to (but not confined) ASUS who are seen as a halo brand with ROG and need to maintain that position and justify the premium they charge.AMDs failure was in trusting the Mobo makers to be sensible rather than implementing a hard cap which they have now done, perhaps the refresh later this year can take the voltage without error we will find out later i guess but for now just update the Bios
Oxford Guy - Saturday, April 29, 2023 - link
AMD didn’t learn from previous iterations of motherboard maker shenanigans, like ASRock 970 boards — ostensibly rated for the 9000-series FX chips that caught fire and ASUS’ penchant (via the Crosshair Formula Z) for underreporting the vcore.Of course, AMD should have been tarred and feathered for the 9000 series in the first place.
abufrejoval - Sunday, April 30, 2023 - link
For me this is AMD's FDIV-bug moment: they either come out now and promise "life-time" replacement warranties, or there will be far too many current owners wondering if they should now burn up their potentially damaged CPU to get a known-good replacement.And let's get rid of that silly notion that PBO/EXPO activation kills warranties while they are at it. It's an advertised feature and using that may not affect warranties: you make it safe to use or you leave it out.
Keith63 - Saturday, May 6, 2023 - link
I have to say nothing really excites me about this motherboard chipset. They obviously use way to much current/wattage which is what is causing this problem. Maybe instead of simply making things faster they should focus on making them energy efficient. I could say the same for graphic cards. When we approach using over 500 watts of constant power just to feed the beast I say we have gone too far. I wouldn't want the monthly electric bill.Shlong - Saturday, May 6, 2023 - link
A few days ago, my 7950X unexpectedly failed. I had kept all settings on Auto, except for enabling XMP DDR5-6000. Initially, I experienced random reboots, but they became more frequent over the past week. Eventually, the system failed to boot into Windows, resulting in a blue screen.
I suspected that the issue might be related to my Windows installation or M.2 drive, so I attempted to boot from a USB Windows Installer. However, it would get stuck right after the POST. I tried various USB bootable media, but only MEMTEST86 would launch. Running a test revealed thousands of errors before crashing.
To troubleshoot, I replaced the 7950X with a 7600X from my secondary system and everything worked flawlessly. When I put the 7950X back into the secondary system, it exhibited the same issues, unable to boot into Windows or USB media. Upon inspection, I noticed some darker gold markers beneath the 7950X where the memory controller is located.
In my 30 years of building PCs, this is the first time a CPU has failed on me. As the push for smaller nanometers continues, AMD should exercise greater caution with voltages as the CPUs seem to be a lot more fragile.
Shlong - Saturday, May 6, 2023 - link
I also submitted an RMA request 3 days ago, but AMD hasn't responded to it yet.martixy - Wednesday, May 10, 2023 - link
Meanwhile GamersNexus released their excellent failure analysis video.Dr_b_ - Thursday, May 18, 2023 - link
"For those who dare" to burn up their CPU and motherboard