AMD RX6000 series cracked dies and blown voltage rails after December driver update?

[Freddie1980]

What statement does he need to correct?

Come one don't be so naive, he quite literally put a video a couple of days ago titled "AMD has a whole new issue to worry about..." and spends most of the video talking about the correlation between the driver update and the cracked die. He also goes on says the the driver update may have updated the vbios which is impossible. He also has nearly 4m subs so anything he puts out will reach a large audience so it's only right he gives everyone the full story.

I'm not sure what he scheduling is like but had he or one of his employees done just a bit of digging before publishing the video it would have clear this relates to a set of GPU's sold by a local miner, he almost suggests this in his video saying it's strange it's only from one repair shop in Germany and there's been no reports from other places around the globe (come on Jay put 2 and 2 together you get what?).

 

[Harlequin]

Make stuff up more please, like you did earlier in the thread.



Good thing you don't do any thing close to journalism.

Yeah cause i was right. Bless you son.

 

[uksoldierboy]

It really does seem that some individuals on here must have been rejected by Lisa at some point.

I can't fathom any other reason why they post the crap they do, does Jensen bring you chocolates or something?

If this was a genuine issue I'd be jumping on the train, whether that's AMD or Nvidia, but from everything I've read, and my own experiences of owning a highly overclocked 6900 with the latest drivers, and a load of users on these very forums having the same experiences..this story looks like utter bull%@#!..

 

[Gerard]

It really does seem that some individuals on here must have been rejected by Lisa at some point.

I can't fathom any other reason why they post the crap they do, does Jensen bring you chocolates or something?

If this was a genuine issue I'd be jumping on the train, whether that's AMD or Nvidia, but from everything I've read, and my own experiences of owning a highly overclocked 6900 with the latest drivers, and a load of users on these very forums having the same experiences..this story looks like utter bull%@#!..

This ^^^

 

[Grim5]

The issue has already been solved, calm down guys

The repair guy managed to dig out what happened from the owners.

Turns out that while all GPU had the same failure mechanism and used the same drivers; what they didn't know and now they do is that all owners had recently purchased the 6800/6900xt and all the purchases were done online from the same seller and this seller had been using the GPUs to mine with.

This is also why the GPUs ended up at a repair shop and not RMA with AMD, the seller did not provide receipts with the GPU and in any case mining voids the warranty on any gpu

 

[Gerard]

The issue has already been solved, calm down guys

The repair guy managed to dig out what happened from the owners.

Turns out that while all GPU had the same failure mechanism and used the same drivers; what they didn't know and now they do is that all owners had recently purchased the 6800/6900xt and all the purchases were done online from the same seller and this seller had been using the GPUs to mine with.

This is also why the GPUs ended up at a repair shop and not RMA with AMD, the seller did not provide receipts with the GPU and in any case mining voids the warranty on any gpu

All this was basically theorised earlier in the thread as well, but noooooo it was a bogey AMD driver to blame for some people!!

 

[Murphy]

Well i learnt something from it, that silicon dies are slightly hygroscopic. I would've though being made of sand and various metals would've made it effectively resistant but i guess tiny amounts of moisture finds it's way in given the right conditions.

 

[Harlequin]

All this was basically theorised earlier in the thread as well, but noooooo it was a bogey AMD driver to blame for some people!!

You mean like the OP

 

[Firegod]

All this was basically theorised earlier in the thread as well, but noooooo it was a bogey AMD driver to blame for some people!!

Yep I think as soon as I saw this, my first comment was "probably ex crypto cards".

 

[Grim5]

We still don't know what the miner did to these cards

Mining shouldn't just kill a whole batch like that nor should it cause all cards to fail at once

The miner may have done something stupid before selling the cards and reassembling.

Maybe the cards were cleaned with a water blaster or something - it has been seen before

 

[muon]

We still don't know what the miner did to these cards

Mining shouldn't just kill a whole batch like that nor should it cause all cards to fail at once

The miner may have done something stupid before selling the cards and reassembling.

Maybe the cards were cleaned with a water blaster or something - it has been seen before

There would be signs of that. Also in the video it was explained some of the cards had been never opened up.

I reckon what is more likely (as hypothesised in the video) is that they were improperly stored for a long time and as they went through heat cycles, condensation formed.

The packaging around the chip can be porous unless glass/ceramic is used.

Moisture sensitivity level - Wikipedia

en.wikipedia.org

 

[Harlequin]

Not sure which video u watched, but krisfix say nothing about that. Remember in every single brand new card is a bag of silica gel, to absorb moisture.

 

[muon]

Not sure which video u watched, but krisfix say nothing about that. Remember in every single brand new card is a bag of silica gel, to absorb moisture.

His most recent one (there are only two, not difficult). He speaks exactly about this.

He even mentions this is something he has learnt from practices when soldering chips. Which is exactly what that initially relates to.

Do you even watch the videos before posting about them?

edit:

I'll provide a link (with a time stamp), since you obviously haven't watched the video.

HERE is what we found, 150 hours of testing - AMD 6000 Series #krisfixgermany #gpurepair

Need a repair? https://krisfix.de/en/geraet-einsendenKrisFix Shop: https://www.gpufix.deQuestion about repair: [email protected] me on Instagr...

youtu.be

 

[Harlequin]

Your interpretation is wildly different than everyone else on the internet, to the point where you are pushing an agenda. Silca gel packs in new gpus absorb moisture.

 

[muon]

Your interpretation is wildly different than everyone else on the internet, to the point where you are pushing an agenda. Silca gel packs in new gpus absorb moisture.

Wow, digging a hole. What silica packs. How do you know how the person selling was storing them. What KrisFix says is very clear, there is no interpretation needed. He talks about temperature and humidity in storing. He talks about when he receives new chips where he doesn't know what conditions they went through, he has to go through a drying procedure. He says if you don't the chips will pop, and that is possibly what has happened here as they reached 80C-100C load.

You are now just being obtuse now that I've provided exactly where he says all of that.

 

[Harlequin]

Yes you had better stop digging that hole. Every single new gpu comes with a silica gel pack , from the factory.

 

[muon]

Yes you had better stop digging that hole. Every single new gpu comes with a silica gel pack , from the factory.

Wow, you literally haven't watched the video. Who says these cards came from the factory?

 

[Mercurial]

Krisfix does say that specifically the coolers are "brand new" with absolutely no dust present, as well as the chips being "soaked" which he theorises is down to storage.

He doesn't offer a theory on why the coolers are so clean from my viewing of the video. It's all very odd.

 

[muon]

Krisfix does say that specifically the coolers are "brand new" with absolutely no dust present, as well as the chips being "soaked" which he theorises is down to storage.

He doesn't offer a theory on why the coolers are so clean from my viewing of the video. It's all very odd.

That is the one gap he doesn't try to answer. But the simplest explanation would be they actually are simply fairly new (or even unused). When ethereum went PoS, there was nothing to do without losing money and so went into storage.

 

[Mercurial]

That is the one gap he doesn't try to answer. But the simplest explanation would be they actually are simply fairly new (or even unused). When ethereum went PoS, there was nothing to do without losing money and so went into storage.

That makes sense. Either way it seems to be a non-issue for the wider install base.

Also I've learnt something though about storage of unused equipment