AMD Releases Second Official Statement Regarding Ryzen 7000X3D Issues

[tabascosauz]

Damn, I was expecting a complicated explanation, but not absolutely all the issues surfacing at once.

Steve's comment about AM5 being a mess really sums up everything pretty well. Everybody is lazy, everybody is incompetent. AGESA is forever a cursed bastard child and AMD will never honestly and transparently communicate with either its users or board vendors, board vendors won't ever listen to user feedback or do better than half-functional BIOSes that don't implement AGESA properly and don't know what OCP and OTP are.........lol @ Asus releasing a first wave of "fixed" BIOSes that didn't fix anything, still kinda funny that Asus chose to adopt Gigabyte's overvolting reputation

That worse silicon quality chips are more likely/sooner to experience damage is in line with any core or IO degradation claims from any past Ryzen generation. The golden samples are never able to replicate early production degradation concerns, and reports would gradually fade as yields improved further into the production run as everyone collectively forgot about it.

Most interesting part of this is the block diagram that shows VDDCR_GFX deriving from Vcore. Skatterbencher was wrong about the topology then (but understandably so). VDDCR_GFX have always taken from VSOC so this not only explains the link of both CCD and IOD exploding, but is a first for AMD. But having iGPU on IO die is also a first, so maybe future APUs will still stick with SOC (which is not preferable).

Looking forward to the video detailing all the platform bugs, anything to give AMD and the board partners a fresh motivational kick in the rear

 

[HD64G]

In conclusion, that is boards' fault due to the bad UEFI implementation from the vendors. The flowed-buggy UEFIs damage the socket and the CPU. If agesa code or silicon were on of the root cause of that fail we would have 90% of the CPU burnt.

 

[Dirt Chip]

In conclusion, that is boards' fault due to the bad UEFI implementation from the vendors. The flowed-buggy UEFIs damage the socket and the CPU. If agesa code or silicon were on of the root cause of that fail we would have 90% of the CPU burnt.

Vendors are for sure responsibl, but AMD can and need to make sure that their CPU function under normal conditions with all it's vendors.
Having multiple vendors settings wrong\damaging values show AMD didn't do enough to inforce the basic that prevent that kind of CPU damage.
The damage to the Asus board, that's entirely on Asus fault though.
And it's in both party interest to prevent those kind of incidence so the responsibility is on both hands, imo at least.

 

[AusWolf]

View attachment 293802
View attachment 293801

So basically, Asus motherboards kill these chips by giving them too much SoC voltage when EXPO is enabled, right?

I'm gonna watch the video now.

 

[Zubasa]

So basically, Asus motherboards kill these chips by giving them too much SoC voltage when EXPO is enabled, right?

I'm gonna watch the video now.

Asus is the worse offender when it comes to sending a crap ton of auto-voltage. Asus boards with non-funnctional OCP is the icing on the cake.
But the other boards vendor are not flawless.
AMD could also do a better job at slapping some sense into the board vendors as well.

 

[AusWolf]

Asus is the worse offender when it comes to sending a crap ton of auto-voltage. Asus boards with non-funnctional OCP is the icing on the cake.
But the other boards vendor are not flawless.
AMD could also do a better job at slapping some sense into the board vendors as well.

And by letting us know what is and isn't an acceptable SoC voltage. More than half a year after release and we still don't know.

It's also a mystery why my (MSi) board sets it to 1.35 V when I enable EXPO, but when I lower it to 1.1 V and then set it back to Auto, it's suddenly 1.2 V (and rock stable).

Asus is usually my go-to brand for motherboards, but I would have had to wait a couple of weeks for the TUF B650M-Plus Wifi, so I went with MSi this time with next day delivery. Now I'm glad I did.

 

[Zubasa]

Asus is usually my go-to brand for motherboards, but I would have had to wait a couple of weeks for the TUF B650M-Plus Wifi, so I went with MSi this time with next day delivery. Now I'm glad I did.

In my experience, Asus' track record with AMD products has always been kind of dodgy.
From GPUs with coolers falling off the card, and AMD laptops with blocked off vents. Now we get this.

 

[R0H1T]

And by letting us know what is and isn't an acceptable SoC voltage.

Tbf that depends on the chip & every chip varies, sometimes a lot. This is why generally even with an OC I use a negative offset on Vcore or SoC because boards more often than not will feed excess voltage. It's also generally better to be on the safer side feeding less voltage even on stable OC IMO.

I use adaptive+negative offset on x570 btw.

 

[AusWolf]

Tbf that depends on the chip & every chip varies, sometimes a lot. This is why generally even with an OC I use a negative offset on Vcore or SoC because boards more often than not will feed excess voltage. It's also generally better to be on the safer side feeding less voltage even on stable OC IMO.

It still wouldn't cost them anything to issue a statement "hey guys, watch your SoC voltage, it shouldn't go above X".

 

[Zubasa]

It still wouldn't cost them anything to issue a statement "hey guys, watch your SoC voltage, it shouldn't go above X".

Yeah and on the same line, it would be great for Intel to publish the max safe VCCSA / System Agent as well. That is basically Intel's equivilent to Vsoc.
There is a lot of conjecture and anectdotes out there but nothing concrete.

 

[R0H1T]

You mean AMD? They should do that because they're probably paying the most for this, aside from the end user of course. Board makers won't do it because they just want to outdo each other in this useless drag race!

 

[Mussels]

I suspect that there are many users who just don't upgrade their BIOS. Some because they don't want to loose custom settings and/or the hassle of having to reapply them.

It is possible for BIOS updates to be issued through Windows Update, and in this instance I think there is a case for AMD to ask its board partners to do this for all motherboard models.

On social media, especially from the intel users BIOS updates are seen as some terrifying ordeal that will either destroy your motherboard, or your OS

because the average user will have a single setting change (SATA from IDE to AHCI, CSM enabled to disabled) and have a total meltdown that somethings not working and blame everything but themselves for not reading the changelog.

Everything i'm seeing in that GN video nwo that i've had time to watch all of it, shows something i've seen in AM4 and argued with a lot of users here on TPU and on facebook about.

Board makers are not enabling or using safety features. If you don't manually set limits, things go wrong, fast.
The biggest issue i keep seeing is people enabling PBO to reach some setting like curve undervolting because a youtuber said so, and that forcibly enables all the other PBO values, often to an unlimited state that relies entirely on the CPUs hardcoded limits

They had a non overclockable CPU receive 400W in the socket and desolder itself. They had multiple boards throw unsafe SoC voltages into the SoC, as well.
The board that sent 400W in, had no VRM heatsinks - they should have been thermal throttling, but it was disabled on the board.

In all the various threads on TPU about people with stuttering issues, it always turns out they've got a motherboard with budget VRMs, PBO enabled with "auto" settings and the assumption that "motherboard limits" and VRM thermal throttles will actually do their job, when it seems to be entirely upto the CPU resulting in stuttering performance as the CPU throttles when the voltages go out of spec

PBO's got EDC (electrical max) values and TDC (Thermally constrained max) values assuming the VRM's will communicate when they're hot, and lower to a safer limit to run a middle-ground, but because theres no communication happening at all between these components the CPU seems to have to guess what the hell is going on based on the voltage it's receiving and is being forced to throttle itself to minimum clocks, rather than to TDC.

It'd be exactly like intel motherboards changing PL2 to 800W with no time limit if you enabled XMP

 

[AusWolf]

On social media, especially from the intel users BIOS updates are seen as some terrifying ordeal that will either destroy your motherboard, or your OS

because the average user will have a single setting change (SATA from IDE to AHCI, CSM enabled to disabled) and have a total meltdown that somethings not working and blame everything but themselves for not reading the changelog.

Everything i'm seeing in that GN video nwo that i've had time to watch all of it, shows something i've seen in AM4 and argued with a lot of users here on TPU and on facebook about.

Board makers are not enabling or using safety features. If you don't manually set limits, things go wrong, fast.
The biggest issue i keep seeing is people enabling PBO to reach some setting like curve undervolting because a youtuber said so, and that forcibly enables all the other PBO values, often to an unlimited state that relies entirely on the CPUs hardcoded limits

They had a non overclockable CPU receive 400W in the socket and desolder itself. They had multiple boards throw unsafe SoC voltages into the SoC, as well.
The board that sent 400W in, had no VRM heatsinks - they should have been thermal throttling, but it was disabled on the board.

In all the various threads on TPU about people with stuttering issues, it always turns out they've got a motherboard with budget VRMs, PBO enabled with "auto" settings and the assumption that "motherboard limits" and VRM thermal throttles will actually do their job, when it seems to be entirely upto the CPU resulting in stuttering performance as the CPU throttles when the voltages go out of spec

PBO's got EDC (electrical max) values and TDC (Thermally constrained max) values assuming the VRM's will communicate when they're hot, and lower to a safer limit to run a middle-ground, but because theres no communication happening at all between these components the CPU seems to have to guess what the hell is going on based on the voltage it's receiving and is being forced to throttle itself to minimum clocks, rather than to TDC.

It'd be exactly like intel motherboards changing PL2 to 800W with no time limit if you enabled XMP

So AM5 boards turned out to be not only unreasonably expensive, but also lacking safety features needed to keep your CPU alive, either due to manufacturer laziness, or chasing the extra 1% in performance and market share. AMD shouldn't allow this.

 

[HD64G]

So AM5 boards turned out to be not only unreasonably expensive, but also lacking safety features needed to keep your CPU alive, either due to manufacturer laziness, or chasing the extra 1% in performance and market share. AMD shouldn't allow this.

CPU and chipset is in control of AMD, board partners get only AGESA and make UEFIs. By the speed the board vendors published new UEFIs and vanished the old buggy ones, I guess some persuation effort from AMD's side was on effect though.

 

[AusWolf]

Right then... 1.3 V max for the SoC it is! What about the cores? My 7700X cores go as high as 1.465 V in short, bursty loads, like opening a program, even without PBO. Is that normal?

 

[HD64G]

If your CPU's power draw doesn't surpass the PPT limit you are ok (~140W me thinks). Check with OCCT's metrics when at full load testing.

 

[AusWolf]

If your CPU's power draw doesn't surpass the PPT limit you are ok (~140W me thinks). Check with OCCT's metrics when at full load testing.

At full load, it settles to around 1.3-1.35 V. It's only those short, bursty loads that scare me.

 

[HD64G]

Zen CPUs work like that on light loads. Since it consumes a few watts during those, you have no danger of it heating up and being damaged.

 

[AusWolf]

Zen CPUs work like that on light loads. Since it consumes a few watts during those, you have no danger of it heating up and being damaged.

How about the voltage alone? Is that no danger to the CPU?

 

[HD64G]

How about the voltage alone? Is that no danger to the CPU?

Not if A and in turn the W are low. A x V = W and that raises temps that kill CPUs or/and boards. V is nothing by itself.

 

[AusWolf]

Not if A and in turn the W are low. A x V = W and that raises temps that kill CPUs or/and boards. V is nothing by itself.

Good to know, thanks.

 

[Mussels]

So AM5 boards turned out to be not only unreasonably expensive, but also lacking safety features needed to keep your CPU alive, either due to manufacturer laziness, or chasing the extra 1% in performance and market share. AMD shouldn't allow this.

Plenty of AM5 board have those features working just fine - these boards have the hardware and the board makers intentionally chose to disable them

We've seen the same thing on intel boards too, so dont play fanboy and make this about AMD

Just a 12700 non-K, no overclocks, having a 42% performance delta on stock settings because almost none of the boards follow intels PL1 and PL2 guidelines, and either result the CPU or VRM's thermal throttling has been ongoing for years now


No one should be blaming intel or AMD for this crap, it's a result of motherboard makers ignoring CPU manufacturer specs without consequences for a very long time

 

[AusWolf]

Plenty of AM5 board have those features working just fine - these boards have the hardware and the board makers intentionally chose to disable them

We've seen the same thing on intel boards too, so dont play fanboy and make this about AMD
View attachment 294144

Just a 12700 non-K, no overclocks, having a 42% performance delta on stock settings because almost none of the boards follow intels PL1 and PL2 guidelines, and either result the CPU or VRM's thermal throttling has been ongoing for years now


No one should be blaming intel or AMD for this crap, it's a result of motherboard makers ignoring CPU manufacturer specs without consequences for a very long time

I'm not playing fanboy. When board manufacturers started doing the same crap with 10-11th gen Core, I called out Intel for not enforcing stricter requirements, just as I'm calling out AMD now. There should be specs that board manufacturers shouldn't be allowed to deviate from under default / auto settings. Sure, board manufacturers should have the common sense not to use dangerous settings in the first place, but they apparently don't. The second line of defense could be Intel and AMD's enforcement of safe defaults, which is also lacking. It hurts their reputation when their CPUs blow up or cook themselves, even if it's entirely the motherboard's fault. I said this back then and I'm saying it now.

With that said, I have an i7 11700 that I love just as much as I love my R7 7700X. Both are extremely versatile, highly configurable chips, especially in a good board. There's no fanboyism on my part.

 

[Mussels]

The've been doing this for years on AMD and intels side, this is just the first time its blown up CPU's that got publicly investigated

Sure, board manufacturers should have the common sense not to use dangerous settings in the first place

That's not common sense - that's an outright crime in many countries, including australia.
The product must be fit for purpose and safe to use, and having safety features is required to get it certified for sale - having them not active at all, is a serious issue.

To be fair i misread the way your comment was written - I read your complaint about AMD boards as if you thought AMD was the source of the problem.

 

[AusWolf]

The've been doing this for years on AMD and intels side, this is just the first time its blown up CPU's that got publicly investigated

I know. I remember review sites calling out certain Intel boards that enforced factory recommendations by default for being "too restrictive", "too slow" and "lying to the customer", and I just thought those boards should be praised, not the ones that break those recommendations.

That's not common sense - that's an outright crime in many countries, including australia.
The product must be fit for purpose and safe to use, and having safety features is required to get it certified for sale - having them not active at all, is a serious issue.

Wow, that's cool! What are the chances of an Australian class-action lawsuit against manufacturers that break OCP, for example?

To be fair i misread the way your comment was written - I read your complaint about AMD boards as if you thought AMD was the source of the problem.

Nah. I know they're not the source - but they could do something against such things by stricter enforcement of electrical safety features.