Space Lynxthe fact that EXPO might have something to do with this burned chips is wild to me... all the Quality Assurance testing that had to be done to even make EXPO a thing, I just don't understand how this was possible. to my knowledge only happened to ASUS boards though, so I guess that is a good and bad, least it is not widespread

It's impossible for EXPO to be the cause per say. The memory profile itself does not touch SoC voltage, only RAM voltage. Where SoC does come into the equation is that many motherboards tend to increase the SoC voltage when EXPO is enabled. Then again, Buildzoid doubts it could be SoC at all. This is down to the fact that the amount of power required to get substrate bubbling cannot be delivered over the SoC rail.

I'd also point out that Ryzen 7000 series has been out since October 2022 and we are only seeing issues now, which could mean this issue is the result of a recent update.

ZubasaEXPO is still a form of overclocking, different DIY motherboards differs on the setting they run at.
There is quite a range of max voltage different brands apply, although anecdotally Asus tends to be more agressive with auto voltages.

EXPO profiles have a set voltage in the profile and even the more aggressive kits are very safe as far as DRAM voltages go. What you are likely referring to is the SoC voltage set by the motherboard but that has nothing to do with EXPO itself. Each vendor tends to have a slightly different approach to SoC voltage but that's down to the motherboard.

This wouldn't be a case of just aggressive voltage, you'd need vastly too much voltage in order to cause an issue. DerBauer tried all the way up 1.8v on the SoC, which is way past extreme, and did not run into an issue. There's also the problem that the SoC rail cannot provide enough power to cause that sort of damage. There'd need to be some kind of short.

ZubasaBascially both Intel and AMD has it in their warranty terms, that XMP / EXPO is considered OC an may void the warranty.
Of course neither of them actually tries to enforce those terms because it will be a PR disaster.
Most of the time the CS won't even ask, and if they do just play dumb and you be fine.

How many Intel cpu's were fried from enabling XMP? I'm going with none.

Why_MeHow many Intel cpu's were fried from enabling XMP? I'm going with none.

It's essentially impossible for XMP or EXPO to fry your CPU as they work right now. All they are is memory profiles that apply timings and a specific memory voltage. The profiles themselves do not set SoC voltage.

Just because you have more CPUs encountering the issue with EXPO enabled does not mean EXPO is at fault, correlation does not equal causation. It could very well be an AGESA bug applying too much SoC voltage when EXPO is enabled or simply abnormal defective CPUs.

At the end of the day everyone is guessing and it'd be better to wait than jump the gun.

After nvidia and their 12pin burnt cable, amd thought they couldn't be left behind and had to take action to close the gap with nvidia on this matter.

N3utroAfter nvidia and their 12pin burnt cable, amd thought they couldn't be left behind and had to take action to close the gap with nvidia on this matter.

Close the gap? Not being intelligent enough to properly plug in a gpu > vs fried cpu's after enabling AMD EXPO.

Why_MeHow many Intel cpu's were fried from enabling XMP? I'm going with none.

Such confidence on motherboards never sent too much voltage to an Intel CPU and killing / Degrading it over the whole Existence of XMP.
All of this without a source. Fact of the matter is you never know.
One thing is for sure, if CPUs actually are at any risk from mem OC alone, AMD / Intel would have done something about it long ago.

Bomby569doesn't EXPO include all the relevant settings?

evernessinceIt's essentially impossible for XMP or EXPO to fry your CPU as they work right now. All they are is memory profiles that apply timings and a specific memory voltage. The profiles themselves do not set SoC voltage.

Just because you have more CPUs encountering the issue with EXPO enabled does not mean EXPO is at fault, correlation does not equal causation. It could very well be an AGESA bug applying too much SoC voltage when EXPO is enabled or simply abnormal defective CPUs.

At the end of the day everyone is guessing and it'd be better to wait than jump the gun.

Yeah the main reason that AMD went with EXPO is actually to avoid trademark issues with Intel. XMP is an Intel trademark.
The profiles stored in SPD chip do not contain much, mostly just the primary timings and tRFC etc an voltage of the memory sticks. The motherboard if free to apply whatever preset to the other timmings an voltages.

BTW one oddity with ASUS statement, all the CPUs that died on their boards have burnt Vcore pins not VSOC or imc.

Edit: From GN

Why_MeClose the gap? Not being intelligent enough to properly plug in a gpu > vs fried cpu's after enabling AMD EXPO.

twas a joke relax

You know... if you look at HWInfo there are two SoC voltage settings. CPU VDDCR_SOC Voltage (SVI3 TFN) and CPU VCORE SoC. Which one should we all be looking at?

trparkyYou know... if you look at HWInfo there are two SoC voltage settings. CPU VDDCR_SOC Voltage (SVI3 TFN) and CPU VCORE SoC. Which one should we all be looking at?

As far as software readings goes, the SVI3_TFN should be the most accurate.
That is dies sense reported directly from the CPU on what it is getting, the other can be what the motherboard VRM claims to be sending.

trparkyWhat's base spec of DDR5?

Sorry I didn't see you ask until now. The base JEDEC spec for DDR5 is 4,800 MT/s CL40.

farmertrueSorry I didn't see you ask until now. The base JEDEC spec for DDR5 is 4,800 MT/s CL40.

The CL40 sounds like crap if you ask me. I wonder why they put the original spec at such a shitty number.

trparkyThe CL40 sounds like crap if you ask me. I wonder why they put the original spec at such a shitty number.

It's just the minimum standard for ram and at a much lower voltage than what the 6,000 CL30 kits run. I want to say JEDEC is around 1.1v where most overclocked DDR5 are 1.35v or 1.4v.