Check Your 4090's 12vhpwr Connectors To Make Sure They are in Pristine Condition

[Sc00p007]

The funny thing is just how many failure points there are; it's not just one issue, the 12hwpr has a whole range of issues that can individually or in combination cause failure

I wonder when 6 and 8 pin was introduced several decades ago; did they have these teething issues as well?

Daisy chaining was the main reason I remember, and the occasional not seated properly. Imo if we take the survey as representative of reality, the difference between 3.3 and 4% I don't think warrants enough of a reaction from 3.3 is fine to 4 omg what a bunch of whoppers, etc. Although I suspect real world numbers will be lower for both.

 

[Aretak]

Just watching that video now. I don't see why it had to change, but Nvidia thought otherwise the absolute whoppers.

I don't think Nvidia really shoulders much blame over all this (for once). The idea of introducing a new connector to serve the needs of power-hungry modern cards isn't a bad one. Frankly, it's silly having to run three or four seperate 8-pin cables to higher-end cards. It's just that the resulting design that's a complete mess thanks to a lack of proper standards from PCI-SIG and manufacturers cutting corners to save half a cent.

 

[Twinz]

Gamers nexus has done a new survey of 23k users

What they found is 12hwpr is 22% more likely to fail than 6 and 8 pin PCIE

I thought they mentioned that most of the failures with the old (6/8 pin PCI-E) connectors involved daisy-chained setups. That's an extra level of stress that none of the 12VHPWR connectors have had to deal with.

So, we are comparing failure rates of old connectors that failed while doing double-duty to 12VHPWR connectors that, as far as I know, have never been daisy-chained at the GPU (failure) side like that.

It's not apples to apples, if that's the case.

The 12VHPWR cable is designed within an inch of its life, with nowhere near the margin of error the old connectors had. To make matters worse, the clip allows "slop" on the opposite side of the connector, creating a situation where a fully-locked connector can *still* end up with a big enough gap to be concerning.

The old connectors also allowed for gaps opposite the locking mechanism but they have so much headroom that the gap is far less likely to cause a problem in the wild.

The updated 12VHPWR connectors should fail less than the originals, or at the very least, fail "better" in that the mode of failure should be failure to power up rather than melting.