AMD Ryzen 7 7800X3D CPU Burns Up

[humbug]

SVI3 is AMD's internal telemetry.

It's simple ohms law. Socket and power plane losses...I've seen this mistake made many times by the written media, no need to delve deeper or try to spin it.

Whose software is reading that and choosing what to do with that information? Its the same argument that a couple of others keep using "But AMD control everything" if you're not using AMD's software, IE their own AGESA and nothing else... no they don't.

 

[Troezar]

SVI3 is AMD's internal telemetry.

It's simple ohms law. Socket and power plane losses...I've seen this mistake made many times by the written media, no need to delve deeper or try to spin it.

So simply put what is measured on the board is not what the CPU gets, there are some losses by the time it gets to the CPU?

 

[Silent_Scone]

Whose software is reading that and choosing what to do with that information? Its the same argument that a couple of others keep using "But AMD control everything" if you're not using AMD's software, IE their own AGESA and nothing else... no they don't.

If ASUS wanted to manipulate voltages wouldn't it be in their interest to do so for the Super IO so that it reads lower?

Nothing you're saying makes sense, so best to conclude here. If you own a 7000 series chip, you can use Ryzen Master to monitor SVI3 telemetry.

So simply put what is measured on the board is not what the CPU gets, there are some losses by the time it gets to the CPU?

Precisely that. GN provided high-side readings as if that was what the CPU was getting and it wasn't. The only reason there would be to show these is if he was educating his audience on calculating power plane resistance, which he wasn't. He did not account for this or the amount of mohms of resistance at the MLCC caps, either. I will add that measuring here isn't unheard of as this is where many have done so in the past, but it's still not enough. Especially when you're claiming the board isn't working.

 

[MartinPrince]

At fault for what ? what CPU overclock ?

you're just making random stuff up now,

Who is at fault for the Actual problem here, not some made up hypothetical, that voltages were set too high on many brands of partner boards,

if AMD gave instructions to the board partners that were wrong. AMD.

If AMD gave instructions to the board partners that were correct and the board partners screwed up implementing them, then the board partners but also AMD for not having systems in place to check their instruction were being followed. business 101 when outsourcing anything.

Save your breath, you're dealing with folks that have AMD rose tinted glasses on so often don't see the wood for the trees where AMD is concerned.

Of course AMD are the top of the tree, i.e. the Dons or the head of family to use mafia parlance. The board partners are like Capos or made men, they take their lead from the head. Asus, Gigabyte, MSI etc make motherboards FOR AMD CPUs not the other way around.

AMD do shoulder some responsibility because their original allowable figure for SOC I believe was 1.4v. If AMD had done some testing prior and discovered that 1.4v was too high for their CPU's and informed the motherboard makers that they must under no circumstance go over x voltage then you would have heard about it or we wouldn't be where we are now.

 

[humbug]

Define too high?

For you and me on a daily use machine, yes it is, for someone with an LN2 pot? no its not.

Have we got to the stage now where the enthusiast PC space is so dumbed down we need people like AMD to hold the hands of Motherboard vendors at every ###### turn because the people they now employ don't know how to read let alone have any clue about electrical engineering?
are we now going to ask AMD to lock all of this crap down to "safe limits" because people like Asus want to cheap out on employing 19 year old collage graduates?

Or am i just getting old?

 

[MartinPrince]

Define too high?

For you and me on a daily use machine, yes it is, for someone with an LN2 pot? no its not.

Have we got to the stage now where the enthusiast PC space is so dumbed down we need people like AMD to hold the hands of Motherboard vendors at every ###### turn because the people they now employ don't know how to read let alone have any clue about electrical engineering?
are we now going to ask AMD to lock all of this crap down to "safe limits" because people like Asus want to cheap out on employing 19 year old collage graduates?

Or am i just getting old?

Sorry to break it to you but yes, you are getting old! We all are but that's not the point. The objective point I think others are making is that AMD does bear some responsibility and you seem to be saying they don't bear any at all. So one has to ask why are you running interference for them? If I remember correctly you did start out by saying that this was ONLY an issue with Asus which has been proved not to be the case and I have not read where you've acknowledged that initial mistake etc.

 

[humbug]

Sorry to break it to you but yes, you are getting old! We all are but that's not the point. The objective point I think others are making is that AMD does bear some responsibility and you seem to be saying they don't bear any at all. So one has to ask why are you running interference for them? If I remember correctly you did start out by saying that this was ONLY an issue with Asus which has been proved not to be the case and I have not read where you've acknowledged that initial mistake etc.

Have other motherboard's popped CPU's?

Yes all of them are culpable to some extent, as for AMD, this issue was not present in their firmware, Steve found one voltage setting that was slightly too low, but not this.

 

[MartinPrince]

Have other motherboard's popped CPU's?

You missed DerBauer's damaged one on a Gigabyte board then?

 

[Sythe79]

AMD's platform? AMD don't make a platform, they make a chip and the firmware to make it work, Motherboard vendors making their own firmware is the responsibility of that motherboard vendors Q&A, not AMD's, its not their product, AMD can't police a third parties software, its not theirs. all they can do is give guidance.

Just like its not AMD's responsibility to police Windows, that's on Microsoft.

https://www.amd.com/en/partner/amd-socket-am5-the-platform-set-to-change-gaming

I think you need to write a strongly worded letter telling AMD they are wrong and you're right.

But by the by problem we suspect was with memory settings which according to you have nothing to do with AMD right let's take a look

https://www.amd.com/en/technologies/expo

OH AMD expo

"AMD EXPO™ technology is a transparent and user-friendly approach to memory profiles. Our rich memory testing documents ensure you can verify hardware compatibility, detailed settings, and complete memory specs, making it easier for you to find the right RAM."

Rich Testing documents ? why it's nothing to do with them, maybe ASUS paid up ?

AMD Ryzen™ Processor Overclocked Memory Compatibility List​

AMD and our memory partners have worked together to create the AMD Ryzen™ processor memory compatibility list. If a kit is included in our list, it has passed a battery of tests on multiple systems as verified by the memory manufacturer, as capable of running at the speed and latency as listed on the box –whether it’s compatible to the XMP standard, or AMD’s new EXPO overclocking memory standard for DDR5.

Why would AMD have memory partners ? nothing to do with them right it's not their non-existent platform.

Nowhere on the page does it imply memory problems lolz not our problem ask maybe Microsoft

 

[Rawr]

If ASUS wanted to manipulate voltages wouldn't it be in their interest to do so for the Super IO so that it reads lower?

Nothing you're saying makes sense, so best to conclude here. If you own a 7000 series chip, you can use Ryzen Master to monitor SVI3 telemetry.




Precisely that. GN provided high-side readings as if that was what the CPU was getting and it wasn't. The only reason there would be to show these is if he was educating his audience on calculating power plane resistance, which he wasn't. He did not account for this or the amount of mohms of resistance at the MLCC caps, either. I will add that measuring here isn't unheard of as this is where many have done so in the past, but it's still not enough. Especially when you're claiming the board isn't working.

To be fair, GN pointed out in their follow up video the SVI3 telemetry was also showing high readings above 1.35v as well.

Now I do think Amd do share blame here though because all board partners were pushing higher than 1.3v. If all of them are then I think it's a higher chance that Amd told them incorrectly than them all being incompetent and overvolting.

 

[humbug]

https://www.amd.com/en/partner/amd-socket-am5-the-platform-set-to-change-gaming

I think you need to write a strongly worded letter telling AMD they are wrong and you're right.

But by the by problem we suspect was with memory settings which according to you have nothing to do with AMD right let's take a look

https://www.amd.com/en/technologies/expo

OH AMD expo

"AMD EXPO™ technology is a transparent and user-friendly approach to memory profiles. Our rich memory testing documents ensure you can verify hardware compatibility, detailed settings, and complete memory specs, making it easier for you to find the right RAM."

Rich Testing documents ? why it's nothing to do with them, maybe ASUS paid up ?

AMD Ryzen™ Processor Overclocked Memory Compatibility List​

AMD and our memory partners have worked together to create the AMD Ryzen™ processor memory compatibility list. If a kit is included in our list, it has passed a battery of tests on multiple systems as verified by the memory manufacturer, as capable of running at the speed and latency as listed on the box –whether it’s compatible to the XMP standard, or AMD’s new EXPO overclocking memory standard for DDR5.

Why would AMD have memory partners ? nothing to do with them right it's not their non-existent platform.

Nowhere on the page does it imply memory problems lolz not our problem ask maybe Microsoft

What's your point here?

 

[Silent_Scone]

To be fair, GN pointed out in their follow up video the SVI3 telemetry was also showing high readings above 1.35v as well.

Now I do think Amd do share blame here though because all board partners were pushing higher than 1.3v. If all of them are then I think it's a higher chance that Amd told them incorrectly than them all being incompetent and overvolting.

SVI3 is as close as it gets. If we want verification on the accuracy of AMD’s own telemetry they would need to weigh in.

What you’re probably referencing is pre-patch where 1.35v SOC was possible. If you watch Bing’s video you will see how to correctly probe the differential sensing pins that are directly connected to the CPU from the VRM controller. These are placed in such a way that the accuracy is very close, and has been present for a few gens now on select boards.

The fact the SIO on these boards is able to closely follow or match SVI3 should be quite telling.

 

[Sythe79]

What's your point here?

I am truly shocked honestly you don't get it, it's OK I didn't expect you to grasp it.

 

[Brizzles]

as for AMD, this issue was not present in their firmware

AGESA 1.0.0.7a ( the latest ) contradicts you. Released on 9th May - way after the CPU burning issue was first raised.

it clearly states it reduces SOC FROM 1.4v to 1.3v

 

[humbug]

AGESA 1.0.0.7a ( the latest ) contradicts you. Released on 9th May - way after the CPU burning issue was first raised.

it clearly states it reduces SOC FROM 1.4v to 1.3v

I'll address the 1.4c thing as well, AMD allowed 1.4v to the SoC, that was the limit cap, they did this so extreme overclockers could have some fun with it. that's not to say Asus or you should set that, unless you're doing LN2 overclocking, the AMD AGESA never did.

AMD have now reduced that cap to 1.3v, because if you can't play nice with your toys we will take your toys away, refer my point in post #1,188.

Define too high?

For you and me on a daily use machine, yes it is, for someone with an LN2 pot? no its not.

Have we got to the stage now where the enthusiast PC space is so dumbed down we need people like AMD to hold the hands of Motherboard vendors at every ###### turn because the people they now employ don't know how to read let alone have any clue about electrical engineering?
are we now going to ask AMD to lock all of this crap down to "safe limits" because people like Asus want to cheap out on employing 19 year old collage graduates?

Or am i just getting old?

 

[Silent_Scone]

hehe, posturing of epic proportions

 

[Rawr]

SVI3 is as close as it gets. If we want verification on the accuracy of AMD’s own telemetry they would need to weigh in.

What you’re probably referencing is pre-patch where 1.35v SOC was possible. If you watch Bing’s video you will see how to correctly probe the differential sensing pins that are directly connected to the CPU from the VRM controller. These are placed in such a way that the accuracy is very close, and has been present for a few gens now on select boards.

The fact the SIO on these boards is able to closely follow or match SVI3 should be quite telling.

I'm on about prepatch yes. The SVI3 for Asus was showing up to 1.4.

 

[Silent_Scone]

I'm on about prepatch yes. The SVI3 for Asus was showing up to 1.4.

Probably. Also higher, given they set 1.45v to try and instigate a failure. This also makes sense as he mentions 1.5v and falsely attributes it to leakage when it was simply him measuring incorrectly from the MLCC caps

 

[Rawr]

Probably. Also higher, given they set 1.45v to try and instigate a failure. This also makes sense as he mentions 1.5v and falsely attributes it to leakage when it was simply him measuring incorrectly from the MLCC caps

No, I mean from EXPO. The information isn't really just from them either there was images from users of it showing it too.

Though as already mentioned, AMD likely told them all that up to 1.4v was safe, so they did.

 

[Silent_Scone]

No, I mean from EXPO. The information isn't really just from them either there was images from users of it showing it too.

Though as already mentioned, AMD likely told them all that up to 1.4v was safe, so they did.

VSOC uses auto rules. Not sure if there are any kits in the wild that have the values programmed into SPD, but it doesn’t change the outcome of what I said.