On the Intel side, XMP controls only memory speed and voltage. EXPO is supposed to operate that way as well. The problem was that core VSOC was so high that it was frying the chip from the inside out, literally. In the case of the faulty Gigabyte and especially ASUS BIOSes, not only was VSOC too high, but the overcurrent protection (OCP) threshold was so high that it might as well have not existed. Steve Burke and his team over at Gamers Nexus demonstrated that even though the CPU itself was obviously already dead, the motherboard was still passing a “power good” signal and continuing to shunt upwards of 400W through the socket, ultimately leading to the fiery endings that were highlighted.
The CPU has substrate, inside that substrate are copper nanotubes to bridge silicon and LGA sections. Those nanotubes were melted and the temperature so high at the ends of them that they actually cavitated the surrounding area underneath the outer wall of the CPU package, shown in cross sections from the failure analysis lab. The melting point of copper is 1984°F / 1084.5°C. That means it got far, far hotter inside there, likely owing to two things: overcurrent, and when the nanotubing failed, a break in the circuit was created, allowing for electrical arcing, which can reach temps of 3000°F / 1649°C.
There was also the VDDIO value, which was higher than it should have been as well. Keep in mind that if VDDIO causes damage, your RAM is also likely shot, especially if EXPO is enabled. All of this should have been communicated by AMD directly to vendors to ensure their BIOS releases never exceeded those voltages. ASUS had an additional failure point though in that in at least one of their BIOS releases, even manually entering a value could cause problems because the “reset to defaults” did nothing at all, meaning your manually entered value was retained even when resetting to defaults via the menu.
It took all the way up until this week for ASUS to even begin to acquiesce and reword their warranty support to not make it so that users who used their beta BIOS wouldn’t just lose their warranty as well. Originally their wording was such that “here is the fix to the problem. Oh, and by the way, if you use this fix you void your warranty”.
AM5 has been an outright mess across all vendors, not just Gigabyte and ASUS, though those two, especially ASUS, was egregiously negligent toward their customers during this whole bruehaha. Which is a shame since the 7800X3D CPUs are very, very good for gaming and in most games should beat out the non-X3D version of the CPU by a tangibly noticeable margin. Unfortunately even now the BIOSes are still iffy and quite frankly I can’t rightly recommend an AM5 X3D CPU due to all of the nonsense.
It isn’t all “sky is falling” now that things are starting to head where they need to, but for the short term at least, I’d hold a “wait and see” approach to AM5. AMD has a lot of reputation to reclaim for not spearheading the effort toward getting this under control before it ended up as clickbait fodder.
As noted above, if you ever need to reset to defaults, double check your values and/or use a multimeter to probe the points on the motherboard to verify the values are within margin of error to what the software is reporting. At least one of their BIOS releases had a non-funcitonal F10 (reset to defaults) function.