As many of you know, we have been working with AMD & Board Partners on Linux & IOMMU groupings. We pulled our Threadripper 1950X CPU today for routine maintenance and discovered some fairly severe pad discoloration that we are still investigating.
Thanks for the heads up. I’ll have to check my 1950x when I find the time in a few days. I have Zenith board and at default settings (non-overclocked) with HWiNFO64 I have seen a few random spikes (cpu voltage) of 1.5 it usually stops at 1.48 though. The voltage typically sits at 1.36 or lower
Many times but almost always on old heavily oc CPUs. The 1920x is not showing any discoloration and we have abused the crap out of it… We also have a 4790k that died the other day and it has no signs. It was a crazy voltage and a 4.4ghz oc iirc.
Seen this primarily on old 920/860/etc parts. But never this quickly. It may just be an oxide layer cleanable with alcohol .
I can’t recall ever seeing this on serverish components even older x99 stuff though.
The pads on the 4930k I swapped for a V2 Xeon were also Clean. That was another high overclock. Hmm…
Hmmm… might be interesting - if somewhat theoretical / academic - to find further common factors apart from the obvious, the latter being heat and material / alloy (ok, kinda hard to tell). How about humidity?
Just my underloaded mind twiddling its thumbs, I guess…
I’ll say it again but only because I already said it elsewhere: Just because the baby makes a mess in its diaper doesn’t mean it’s gonna die. That said, something smells funny and something should be dealt with before something gets out of hand. Many things can cause discolouration of terminals. Could it be the result of some sort of cathodic reaction? The effect is so limited one wonders if it might not be due to some micro-contamination. At any rate it might be something worth looking into on AMDs part as it does seem pretty early in the game for this sort of thing to happen. Personally, I’d chalk it up to growing pains. AMD is only beginning to lend itself to this sort of LGA type of design. I don’t suggest they should go back to using the old “pins on the socket” architecture but maybe they need a little time to work it out. Send it back to them to have it ANALyzed and pray they don’t find remnants of Wendelizian Tea in the oxide.
My guess is that excessive current is being pulled from too few pads, causing them to heat up, and cause that severe discolouration. It’s definitely not being caused by prolonged heat; the CPU has just not had enough time, plus, the discolouration is too localized.
Here’s some CPU discolouration on an Intel CPU. Notice it around the edges of the CPU, along with the top right. This appears to be more-so caused by prolonged heat to the CPU, and not just individual contacts getting very hot, as is happening to Wendell’s 1950X.
Wonder how long those CPU’s will work in the long run, if this discolouration is already appearing a few months into the release of thread ripper.
Actually I’m starting to think I just have a really bad board…
one of the mounting inserts for the cooler just popped out and oh right, mem channel A0 and A1 were not working…
welp, time for an RMA. That coloration doesn’t actually look too bad tbh. For the 2990WX I had running PBO under water at 4ghz the whole time. It looked like a baked potato.
I cleaned it and put it back together and… still running… knock on wood… running 24/7 since launch… not died yet. Several power outages and one block flush/deep cleaning.
that discoloration almost looks like b/c of uneven mounting pressure.