Carzy Zarx’s PC Gaming Emporium - Catch Up on All the Latest PC Gaming Related News

[Jizz_Beard_thePirate]

JEMC said:

Captain_Yuri said:

After watching the video, it doesn't exactly spark the highest of confidence going forward... Maybe I'll go with Intel this time around...

I didn't get most of what the video said, the constant back and forth between the different voltages confused me, but here's what I got, please correct me if I'm wrong:

• There is more than one problem with the voltages. In the video he talks about two of them: the one about EXPO and another one related to the voltage going to the IO die, especifically the iGPU.
• AMD send guidelines to its motherboard partners but didn't check or made sure that they followed them.
• All the different partners are at fault, but Asus is the worst.
• The problem with EXPO was that Asus went the easy route and applied unsafe amounts of voltages to ensure that EXPO worked.
• The protection systems that Asus boards should have don't work, even those that Asus advertises to offer greater protection.
• Gigabyte has a problem where it applies more voltage than what the system reports.
• There will be more videos detailing the other problems.

Did I miss something?

There's basically two aspects to the issue. One is the CPU killing itself due to voltages which affects all vendors and the other aspect is the motherboard killing itself instead of protecting itself despite being able to which is mainly Asus but could be with other vendors as well.

The reason why the CPU is killing themselves is something that GN doesn't have a full understanding of yet. Their theory is the failure may be triggered by a differential between SoC and other voltages potentially originating from the IO die around the iGPU area. This could be something that happens asap or it could be something that degrades the cpu overtime. Once the problem happens, it creates a short killing the CPU. At this stage, the Over Current Protection from the motherboard vendors is supposed to kick in and save the motherboard from dying along with the CPU. The problem with Asus motherboards is that the OCP doesn't kick in and instead it continued to feed the dead CPU with more and more power until the dead CPU reaches well over 200C and the pins melt. Asus has a sophisticated controller on the motherboard that's supposed to detect this but Asus set the OCP limits too high so the controller says "400 watts going to at 120 watt CPU? Sure why not?"

AMDs current fix is to limit various sets of voltages to make sure this doesn't happen and not just SoC/Expo. This should lessen the degradation or fully eliminate it but GN believes that there could be more issues because the platform is such a mess as they dug into it deeper. As of right now, if the CPU does fail, AMD will warranty it even under Expo but no one is sure if Mobo will agree to it.