Ryzen Owners Zen Garden

[evernessince]

If VSOC is truly the culprit, I expect the problem to no longer be relevant for brand new CPUs, and that's how AMD will sweep it under the rug. Replace the high-profile failures and maybe the occasionally CPU that is close enough to make it to failure in the next few months, then revise their policy (although their RMA should be lax enough to continue allowing replacements anyway through warranty).

But that's assuming that we know the whole story and can solely pin the blame on VSOC. I don't believe that's the case, only Asus has been obscene with VSOC, but other board vendors have also produced these failures (including those that were already below 1.3V).

I really don't see how only partially addressing the issue would be a net positive for AMD, even just from a financial perspective. You are talking about them hiding an issue that can cost them millions in returns and even more in a potential lawsuit and both liabilities only grow as more people buy into the platform. It doesn't make sense for them to make a move like that, the cost is far greater than the benefit. It makes far more sense for them to spend the pittance in comparison to fully fix the issue and offer RMAs to all potentially impacted CPUs. Even if we assume it's a CPU hardware level issue (which is extremely unlikely), it makes much more sense to RMA the small amount of CPUs on the market now than wait for that number to drastically increase. This issue isn't something that could be easily hidden, if Zen 4 CPUs started dying en mass 3 years out AMD would 100% be on the hook for intentionally selling a defective product and suppressing knowledge of that defect.

But that's assuming that we know the whole story and can solely pin the blame on VSOC. I don't believe that's the case, only Asus has been obscene with VSOC, but other board vendors have also produced these failures (including those that were already below 1.3V).

No, Gigabyte has been caught with vastly inflated voltages as well. Buildzoid did a whole video on how his AM5 gigabyte board was setting his voltage to 1.45v despite him manually setting it to 1.2v.

 

[Space Lynx]

GN video is out. Lots of very cool imagery. Just some thoughts:

1. The acoustic scan result is interesting - the bubbles(?)/voids in the indium solder application could certainly explain some of the wack core deltas that some samples have.
2. Crazy how little thickness the base die and Vcache take up. Maybe the silicon spacer isn't to blame for X3D thermals, as much as cache being in the way of heat dissipation from the CCD.
3. So allegedly the theory goes high VSOC >> exponentially more leakage and degradation, where low SP samples have a head start >> damage is extensive enough to short and deliver massive current in small area >> silicon melts/indium solder melts/heat builds up >> boom takes path of least resistance?

Steve saying that the precursor degradation is irreversible......obviously, everyone knows that's how silicon degradation works, but in past generations "degradation" due to bad SP quality and/or unsafe OC was often visibly obvious. On AM5 there haven't been many consistent, visible signs of this sort of "slow damage" over time. As an existing [Asus] owner, you'd just have to update to a safe BIOS and hope that there's still enough life in the CPU to last for x years..........and hope that the board now has working OCP that can hopefully limit the extent of damage (ie. saving the board) if it does happen. Because AMD sure as hell won't replace AM5 CPUs en masse for preventative reasons.

I understand that the lab is just being honest when they say they can't isolate any of the known factors (thermal runaway, short, VSOC degradation) from dozens of potential manufacturing or environmental factors (just rattled off like, 30+ non-exhaustive examples), but the video hardly arrives at the "Truth About AMD's CPU Failures" - it just leaves more questions than answers.

If anything, GN's investigation confirmed just one thing: AMD won't be the one to deliver that "Truth", pretty clear from their one and only official statement on the matter that they believe the new AGESA voltage limits will make the problem magically go away.

just finished the video, holy shit, that was next level. gamers nexus is legit lol

 

[tabascosauz]

I really don't see how only partially addressing the issue would be a net positive for AMD, even just from a financial perspective. You are talking about them hiding an issue that can cost them millions in returns and even more in a potential lawsuit and both liabilities only grow as more people buy into the platform. It doesn't make sense for them to make a move like that, the cost is far greater than the benefit. It makes far more sense for them to spend the pittance in comparison to fully fix the issue and offer RMAs to all potentially impacted CPUs. Even if we assume it's a CPU hardware level issue (which is extremely unlikely), it makes much more sense to RMA the small amount of CPUs on the market now than wait for that number to drastically increase. This issue isn't something that could be easily hidden, if Zen 4 CPUs started dying en mass 3 years out AMD would 100% be on the hook for intentionally selling a defective product and suppressing knowledge of that defect.

No, Gigabyte has been caught with vastly inflated voltages as well. Buildzoid did a whole video on how his AM5 gigabyte board was setting his voltage to 1.45v despite him manually setting it to 1.2v.

Yes, I saw bz's video on auto-VSOC upon crossing VDIMM threshold. I did not praise Gigabyte, only saying that the failures aren't limited to Asus despite Asus being the standout offender. The way ASRock and MSI are relatively conservative on VSOC should never have allowed a single failure to happen on their boards.

So your position is that AMD's statement is definitive because they always do a cost/benefit analysis? Remember how many tries it took them to get out a BS-free statement on the MBA vapour chamber issue?

wendell loves to make novice-level assumptions in his board reviews, but his point about SP variation makes a lot of sense. It's not as if Ryzen hasn't suffered from bad silicon quality before: launch Matisse parts literally never reaching advertised boost even after AGESA 1004, bad SP launch Vermeer parts either suffering from cache hierarchy/bus interconnect or outright dying for no reason, or the batch of 5800X3Ds around end of last year that gave up the ghost. AMD never publicly "got to the bottom" of any of those, just relied on yields improving over time to sweep itself under the rug. If the issue isn't more complex than the speculated failure route, then clamping down on all the rails will be enough of a "fix".

Like I said, by far the best solution would be for AMD to conduct as thorough an investigation as third parties have, but that's traditionally a tall order for AMD.

Besides, to be fair to AMD, how do you cast a net wide enough to cover "affected" CPUs when there aren't even clearly and consistently discernable signs of damage that we know of? Everyone that bought an Asus board?

 

[Space Lynx]

Yes, I saw bz's video on auto-VSOC upon crossing VDIMM threshold. I did not praise Gigabyte, only saying that the failures aren't limited to Asus despite Asus being the standout offender. The way ASRock and MSI are relatively conservative on VSOC should never have allowed a single failure to happen on their boards.

So your position is that AMD's statement is definitive because they always do a cost/benefit analysis? Remember how many tries it took them to get out a BS-free statement on the MBA vapour chamber issue?

wendell loves to make novice-level assumptions in his board reviews, but his point about SP variation makes a lot of sense. It's not as if Ryzen hasn't suffered from bad silicon quality before: launch Matisse parts literally never reaching advertised boost even after AGESA 1004, bad SP launch Vermeer parts either suffering from cache hierarchy/bus interconnect or outright dying for no reason, or the batch of 5800X3Ds around end of last year that gave up the ghost. AMD never publicly "got to the bottom" of any of those, just relied on yields improving over time to sweep itself under the rug. If the issue isn't more complex than the speculated failure route, then clamping down on all the rails will be enough of a "fix".

Like I said, by far the best solution would be for AMD to conduct as thorough an investigation as third parties have, but that's traditionally a tall order for AMD.

Besides, to be fair to AMD, how do you cast a net wide enough to cover "affected" CPUs when there aren't even clearly and consistently discernable signs of damage that we know of? Everyone that bought an Asus board?

I love my MSI boards, good to know they are still a safe bet even in these turbulent times!

 

[AusWolf]

I love my MSI boards, good to know they are still a safe bet even in these turbulent times!

Good to know that not waiting 2 weeks for that lovely-looking Asus TUF that I originally wanted, and getting the MSi Pro with next day delivery instead was the right choice.

 

[evernessince]

So your position is that AMD's statement is definitive because they always do a cost/benefit analysis

No, my position is what I stated. That it doesn't make good business sense for AMD to hide the issue and only partially address it.

Remember how many tries it took them to get out a BS-free statement on the MBA vapour chamber issue?

According to most news articles, once the issue was escalated beyond AMD's front line service reps it was addressed as a vapor chamber issue.

AMD says insufficient fluid in Radeon RX 7900 XTX vapor chamber causing overheating, offers replacements

Last month brought reports that some users of Radeon RX 7900 XTX MBA (Made By AMD) reference cards were experiencing hotspot temps of up to 110C. A...

www.techspot.com

Every other article I see echos the same sentiment.

Besides, to be fair to AMD, how do you cast a net wide enough to cover "affected" CPUs when there aren't even clearly and consistently discernable signs of damage that we know of? Everyone that bought an Asus board?

Any board that set VSOC above 1.30V automatically with or without EXPO enabled assuming VSOC is the issue. Of course it's up to the customer if they ultimately RMA but it doesn't seem too difficult.

 

[Space Lynx]

No, my position is what I stated. That it doesn't make good business sense for AMD to hide the issue and only partially address it.



According to most news articles, once the issue was escalated beyond AMD's front line service reps it was addressed as a vapor chamber issue.

AMD says insufficient fluid in Radeon RX 7900 XTX vapor chamber causing overheating, offers replacements

Last month brought reports that some users of Radeon RX 7900 XTX MBA (Made By AMD) reference cards were experiencing hotspot temps of up to 110C. A...

www.techspot.com

Every other article I see echos the same sentiment.



Any board that set VSOC above 1.30V automatically with or without EXPO enabled assuming VSOC is the issue. Of course it's up to the customer if they ultimately RMA but it doesn't seem too difficult.

wait for part 3 of Steve's video from gamers nexus.

of course thanks to this video, it appears Asus is already changing its tune...

ASUS Posts an Important Update for Ryzen 7000 Processors on its AM5 Motherboards

This update is a continuation of our response to this issue. As mentioned, ASUS has been working with AMD on defining new rules for the new BIOS updates. These updates will continue to be completed as necessary. The BIOS updates address concerns regarding potential damage that may occur when...

www.techpowerup.com

without steve from gamersnexus... i sometimes wonder where we would be... he keeps them honest.

 

[A Computer Guy]

Any board that set VSOC above 1.30V automatically with or without EXPO enabled assuming VSOC is the issue. Of course it's up to the customer if they ultimately RMA but it doesn't seem too difficult.

In that case it should also be possible to setup a test bed of CPU's with imaging before and after various test run scenarios to see when and to what extent degradation occurs. I wonder if AMD would do that to also give them insights into building quality standards the motherboard vendors should adhere to especially for future chips where the nodes are continuing to get smaller and more refined. It would be good for AMD to avoid, where possible, situations board vendors could end up creating excessive warranty, recall, and brand reputation damage. It also behooves AMD to ensure it doesn't skimp on AGESA care and quality.

wait for part 3 of Steve's video from gamers nexus.

of course thanks to this video, it appears Asus is already changing its tune...

ASUS Posts an Important Update for Ryzen 7000 Processors on its AM5 Motherboards

This update is a continuation of our response to this issue. As mentioned, ASUS has been working with AMD on defining new rules for the new BIOS updates. These updates will continue to be completed as necessary. The BIOS updates address concerns regarding potential damage that may occur when...

www.techpowerup.com

without steve from gamersnexus... i sometimes wonder where we would be... he keeps them honest.

The GN video is not that long but I fell asleep 1/2 way though watching the recent video then later had to rewatch it. I find Steve's voice combined with the rapid number of words per minute is sometimes like listening to relaxing progressive/psytrance music and I get sleepy but wake up amazingly refreshed and alert like 15 minute later. Then again maybe it's information overload.

 

[freeagent]

IMO the truth is AMD rushed this out too fast to compete with Intel with DDR5 memory speeds and on the CPU clock speed scene taking what they learned with Zen3. The only problem is they did it too fast with unproven tech. Why do you think they continue to play it safe with the 58X3D. It is completely hobbled compared to the others in the Zen3 line. It cannot take the same stresses. Now they are adding more stresses on a smaller package using more power than Zen3.

zzzt is a terrible sound

 

[Klemc]

I don't care about overclock, that's even a good thing to me the X3D don't have overcloackability, but now it's a bout EXPO that can't even be enabled, btw that's not that much important but 6000 instead 4800 is good tho

What's the best ?.. 4800 38-38-38 or 5200 40-40-40

FACTORY TIMING PARAMETE
• Default (JEDEC): DDR5-4800 CL40-39-39 @1.1v
• XMP Profile #1: DDR5-6000 CL32-38-38 @1.35v
• XMP Profile #2: DDR5-5600 CL40-40-40 @1.25v
• XMP Profile #3: DDR5-4800 CL38-38-38 @1.1v

 

[A Computer Guy]

I am still considering just saying yolo and getting one myself. I'm not worried at all about soc damage... bios update fixed all that already.

I have to say the turnaround on fixing the issue seemed pretty quick.

 

[Space Lynx]

I have to say the turnaround on fixing the issue seemed pretty quick.
View attachment 295205

lmao

 

[tabascosauz]

In that case it should also be possible to setup a test bed of CPU's with imaging before and after various test run scenarios to see when and to what extent degradation occurs. I wonder if AMD would do that to also give them insights into building quality standards the motherboard vendors should adhere to especially for future chips where the nodes are continuing to get smaller and more refined. It would be good for AMD to avoid, where possible, situations board vendors could end up creating excessive warranty, recall, and brand reputation damage. It also behooves AMD to ensure it doesn't skimp on AGESA care and quality.

The GN video is not that long but I fell asleep 1/2 way though watching the recent video then later had to rewatch it. I find Steve's voice combined with the rapid number of words per minute is sometimes like listening to relaxing progressive/psytrance music and I get sleepy but wake up amazingly refreshed and alert like 15 minute later.

I don't think any of the lab testing has demonstrated what the slow VSOC degradation looks like. Lab is only concerned with the end result looks when so much damage has accrued from electromigration/charge trapping(?) that some other unknown event takes place and nukes the CPU. I'm not sure it's even possible to "see" degradation, that's something that happens on more or less the molecular (atomic?) level.

wendell posited it as a theory, but GN is pretty confident in silicon quality's relationship with the degradation, so even if you had a handful of AM5 CPUs to test it wouldn't mean much - there could conceivably be a CPU with good enough SP that outlasts the test or never dies.

I don't care about overclock, that's even a good thing to me the X3D don't have overcloackability, but now it's a bout EXPO that can't even be enabled, btw that's not that much important but 6000 instead 4800 is good tho

What's the best ?.. 4800 38-38-38 or 5200 40-40-40

FACTORY TIMING PARAMETE
• Default (JEDEC): DDR5-4800 CL40-39-39 @1.1v
• XMP Profile #1: DDR5-6000 CL32-38-38 @1.35v
• XMP Profile #2: DDR5-5600 CL40-40-40 @1.25v
• XMP Profile #3: DDR5-4800 CL38-38-38 @1.1v

Both are shit

Chances are you don't need 1.3V to run 6000, so dropping to JEDEC is a stupid overreaction.

 

[AusWolf]

I don't care about overclock, that's even a good thing to me the X3D don't have overcloackability, but now it's a bout EXPO that can't even be enabled, btw that's not that much important but 6000 instead 4800 is good tho

What's the best ?.. 4800 38-38-38 or 5200 40-40-40

FACTORY TIMING PARAMETE
• Default (JEDEC): DDR5-4800 CL40-39-39 @1.1v
• XMP Profile #1: DDR5-6000 CL32-38-38 @1.35v
• XMP Profile #2: DDR5-5600 CL40-40-40 @1.25v
• XMP Profile #3: DDR5-4800 CL38-38-38 @1.1v

It may sound like blasphemy, but I can't feel any difference between 6000 CL36 and 4800 CL40. I don't even know why I've got EXPO enabled, to be honest.

Both are shit

Chances are you don't need 1.3V to run 6000, so dropping to JEDEC is a stupid overreaction.

That too. My 7700X has been fine with 1.2 VSOC at EXPO right from the start.

 

[Space Lynx]

It may sound like blasphemy, but I can't feel any difference between 6000 CL36 and 4800 CL40. I don't even know why I've got EXPO enabled, to be honest.


That too. My 7700X has been fine with 1.2 VSOC at EXPO right from the start.

im going to try this if i go yolo and get a 7800x3d.

1.2v soc reboot, enable expo, reboot, and then done


why did you add in the extra step of soc auto after manual set of 1.2 when you told me before?

 

[AusWolf]

im going to try this if i go yolo and get a 7800x3d.

1.2v soc reboot, enable expo, reboot, and then done


why did you add in the extra step of soc auto after manual set of 1.2 when you told me before?

My board is weird. Enabling EXPO gives me VSOC = VDIMM = 1.35 V (1.3 V after the update with 1.35 VDIMM). If I manually lower it to 1.1 V, reboot, set it back to Auto, I get 1.2 VSOC.

Don't mention the 7800X3D. I'm so tempted!

 

[Space Lynx]

My board is weird. Enabling EXPO gives me VSOC = VDIMM = 1.35 V (1.3 V after the update with 1.35 VDIMM). If I manually lower it to 1.1 V, reboot, set it back to Auto, I get 1.2 VSOC.

Don't mention the 7800X3D. I'm so tempted!

well i am looking at the same mobo, so this is good to know. yeah mobo's are weird on this kind of stuff, i don't think its just this generation.

 

[AusWolf]

well i am looking at the same mobo, so this is good to know. yeah mobo's are weird on this kind of stuff, i don't think its just this generation.

I was a bit worried at first, considering it's kind of a budget board (as much as any B650 board can be called that), but it's surprisingly nice. Stable, easy to use, even Memory Context Restore works with the latest update, so boot time is minimal. The factory BIOS was crap, though, so if you buy it, update straight away. It didn't boot with EXPO, and gave me blue screens at JEDEC. Now, the only weirdness left is what I mentioned above.

Oh, and one more thing: there's an option to change your processor's TDP values, but it only works downwards for some reason. Higher values don't do anything.

 

[tabascosauz]

It may sound like blasphemy, but I can't feel any difference between 6000 CL36 and 4800 CL40. I don't even know why I've got EXPO enabled, to be honest.

If you only play at 1080p 60Hz, hardware requirements are very, very low. Memory differences more apparent at higher fps.

 

[AusWolf]

If you only play at 1080p 60Hz, hardware requirements are very, very low. Memory differences more apparent at higher fps.

Interesting. I would expect the opposite, considering that CPUs show the greatest differences in performance at lower resolutions.

 

[tabascosauz]

Interesting. I would expect the opposite, considering that CPUs show the greatest differences in performance at lower resolutions.

Yes, but it's the equivalent of a 60Hz framecap, so you don't get the opportunity to see the differences. Benchmarks will still show it though

 

[AusWolf]

Yes, but it's the equivalent of a 60Hz framecap, so you don't get the opportunity to see the differences. Benchmarks will still show it though

Exactly. I only look at CPU reviews as future reference, to be honest.

 

[Space Lynx]

my 5800x3d arrives tomorrow. I am wondering for my clean install, should I use the latest AMD drivers that W1zz has on this site, or amd.com

or should i use this from my motherboard msi page that was posted today... why are the numbers different? 23.4.3 for amd.com and w1zz, and here its 22.40.46

not sure which to use, and frankly, I am still confused why this is even a thing on a motherboard page. shouldn't it just use w.e amd is using at that time, like 23.4.3?

PRO B550-VC

Best AMD AM4 B550 ATX motherboard, lightning PCIe 4.0, lightning M.2, Extended heatsink, USB 3.2 Gen 2, M.2 heatsink, powerful overclocking, MSI PRO

www.msi.com

 

[AusWolf]

my 5800x3d arrives tomorrow. I am wondering for my clean install, should I use the latest AMD drivers that W1zz has on this site, or amd.com

or should i use this from my motherboard msi page that was posted today... why are the numbers different? 23.4.3 for amd.com and w1zz, and here its 22.40.46

not sure which to use, and frankly, I am still confused why this is even a thing on a motherboard page. shouldn't it just use w.e amd is using at that time, like 23.4.3?

View attachment 295355

PRO B550-VC

Best AMD AM4 B550 ATX motherboard, lightning PCIe 4.0, lightning M.2, Extended heatsink, USB 3.2 Gen 2, M.2 heatsink, powerful overclocking, MSI PRO

www.msi.com

You're looking at VGA drivers. Search for the chipset driver on the AMD website.

 

[Space Lynx]

You're looking at VGA drivers. Search for the chipset driver on the AMD website.

chipset is separate from gpu drivers? what is vga? i thought vga was video card?

and gpu

so what is the the vga? and the number doesn't match either of these, even though it was just released today on MSI motherboard website