ASUS Unveils Exclusive PBO Enhancement for AMD X670, B650 Motherboards

[RJARRRPCGP]

The score drop may be a symptom of instability! Even if you don't have a WHEA error.

 

[OneMoar]

That was because of SOI (silicon on isulator) vs Bulk silicon

the only people that used SOI are gloflo/ibm
none of the intel or amd cpus have ever used SOI since 32nm and certainly none in in the last 10 years
and FD-SOI are a related but different thing entirely that just means the transistors them selves use a silicon dioxide as a insulator that has no bearing the operating temp of the chip has a whole
which most phased out by 14nm

 

[Airisom]

Well, the difference between laptops and desktop systems is that desktop processors draw more power than their laptop counterparts, so electromigration at those 90C+ temperatures isn't really an issue in mobile products. You're talking about 95C on a <45W part pulling little current vs 95C on a 5nm 200W part pulling a lot of current.

I'm not sure if your comment on a smaller node = a higher thermal tolerance is correct. Just looking at microprocessors from a high level, you have denser transistors that are physically smaller than older nodes. If the transistors are smaller, then current boundaries should scale with the decreased node size. Pushing the envelope will increase the risk of electromigration if the increased thermal density cannot be cooled adequately - and that's exactly the issue that people are debating now. 5nm is a dense node, the IHS is thick, TDP is up from Zen 3, so the result are high temperatures.

In the end, a good percentage of interested consumers are not comfortable with their processors running that hot, even if it is only a few cores hitting those numbers, and it could have been prevented. AMD could have decreased IHS thickness to aid in more efficient heat transfer, but they opted to maintain AM4 cooler compatibility instead. They could have dropped TDP to 125w and kept 95% of performance while having drastically lower temperatures. My guess is AMD pushed these to the limit because of Raptor Lake, so we'll see how that goes in a few days.

 

[OneMoar]

electromigration isn't a thing anymore modern processes and improvements in transistor design have effectively rendered it null and void unless your cpu happens to be running at 120c+ and sucking down 300A then you should probably be concerned as the atoms are probly getting pretty angry by then

95c pppff 95c is cold from a process standpoint
they are tested at 125c
so worst case scenario boosting to ~5Ghz @ 95c @ 1.25v its still going to last 15 years
and that is run 24/7 @ load


amperage @ temperature is the concern, ryzen 7000 requires significantly less vcore and thus lower current
stress aging data at 0.96 V and 125C on the 5nm FOM ring oscillator made with the High Mobility Channel finFETs shown in Fig. 12 with improved aging relative to the 7nm node.

Fig. 12. Plot showing T50% lifetime(years) vs. stress voltage Vstr of aging study at 125C of N5 HMC finFET ring oscillators and N7 silicon finFET ring oscillators showing improved aging at the 5nm node relative to that at 7nm.

 

[Mussels]

You cant use decades old advice and "common knowledge" on something that's

1. Incredibly advanced microscopic techonology

2. Proprietary tech


All we can know is what comes out from long term use - and laptops and GPUs have had zero issues at these temperatures for many years now. The only issues have been when the COOLING fails.


You can unplug the cooler on any modern CPU and see them hit these temps and lose performance, but they dont exactly crash or take damage and tick a magical counter that says 'slow me down'

 

[OneMoar]

go ask Dr. Ian Cutress
unsubbing this thread serves no point anyway