EVGA Reveals that Bad Soldering Was the Culprit behind Bricked RTX 3090 Cards

[Chrispy_]

what ?!,
go read any GPU reviews on TPU and see the Power consumption on furmark test, up to 20% power increase in furmark than peak gaming, also to your point the EVGA 3090 survived Furmark tests, so clearly the soldering can handle furmark over extreme condition which proves the root of the problems is a misconfigured bios

Gaming is not the only use case for a GPU, and regardless of what the load is, Furmark still doesn't exceed the rated TDP specified by the board manufacturer and what's set on the power limit slider.

If it was a soldering quality, then we would have heard that from many reviews and users who stress tested the cards using stress softwares like Furmark

Uh, EVGA have identified the problem and declared, unambiguously that it IS a soldering quality problem.

Different types of load stress different parts of the GPU; Inductor whine, for example, is a very audible sign of mechanical stress on soldered inductors and that behaviour is common in menus and other high-framerate, low-load scenarios where the GPU is mostly idle. It has crashed/damaged cards before and it will crash/damage cards in the future unless designs accommodate this rather common scenario.

The card design and soldering quality have to meet all possible load scenarios, not just gaming and furmark.

 

[eidairaman1]

The thing that bothers me is they way evga gets special treatment as a company from famous review sites. They refused to name the company and made it sound like every other company had the problem but it was only one, and they knew it. Yet, they manipulated their position.

They are paying those reviewers off

Gaming is not the only use case for a GPU, and regardless of what the load is, Furmark still doesn't exceed the rated TDP specified by the board manufacturer and what's set on the power limit slider.


Uh, EVGA have identified the problem and declared, unambiguously that it IS a soldering quality problem.

Different types of load stress different parts of the GPU; Inductor whine, for example, is a very audible sign of mechanical stress on soldered inductors and that behaviour is common in menus and other high-framerate, low-load scenarios where the GPU is mostly idle. It has crashed/damaged cards before and it will crash/damage cards in the future unless designs accommodate this rather common scenario.

The card design and soldering quality have to meet all possible load scenarios, not just gaming and furmark.

Mining is not graphics acceleration.

 

[Bomby569]

I dont believe them, EVGA RTX 3090 cards have been around long before New World came, and were stressed to the max by users playing a verity of games, its something about the New World game engine the triggers something in the card bios, creating problems.

If it was a soldering quality, then we would have heard that from many reviews and users who stress tested the cards using stress softwares like Furmark

EVGA 3090's have been dying way before NW was even a thing, since launch, the problem is no one talks about it. But on EVGA forums you can hear about people that have been RMA'ing the cards for months now.

This is what is shitty about this all situation, all the media is covering up this thing, no one talks about it. If it were Gigabyte or MSI we wouldn't hear the end of it.

 

[Chrispy_]

Mining is not graphics acceleration.

Well obviously. How is that relevant?

 

[R-T-B]

Mining is not graphics acceleration.

And graphics acceleration is not all a gpu does anymore, even when just gaming.

 

[HeavyHemi]

since the industry has changed the solder to lead-free problems just increased especially at bga contacts of laptop gpu's...

the new lead-free solder doesn't handle so well the thermal expansion and contraction and have more internal flaws in the structure

New? You're 20 years behind the times.

 

[RJARRRPCGP]

FGS, lead-free soldering was the norm, since the socket 775 era! And there are parts still soldering along! A lot of socket 1366 motherboards are trucking along!

 

[chrcoluk]

The AIB build issues are the worst in Ampere I have known for a long time, remember we also now have gimped drivers (old adaptive clock scheduler removed) because of components used that couldn't handle rapid load increases.

 

[Vayra86]

EVGA 3090's have been dying way before NW was even a thing, since launch, the problem is no one talks about it. But on EVGA forums you can hear about people that have been RMA'ing the cards for months now.

This is what is shitty about this all situation, all the media is covering up this thing, no one talks about it. If it were Gigabyte or MSI we wouldn't hear the end of it.

Of course, there is a reason EVGA has that fantastic warranty policy. They know they need it to keep customers' faith after the umpteenth shit cooler.

That, and there is a special type of EVGA customer called 'Elites' that are very elite at sending cards back and forth They paid the premium beforehand, EVGA's fine with it. So on EVGA forums, the problems look like they always exist, even if majority of cards are OK.

The AIB build issues are the worst in Ampere I have known for a long time, remember we also now have gimped drivers (old adaptive clock scheduler removed) because of components used that couldn't handle rapid load increases.

That was the first writing on the wall after we learned we got Samsung 8nm alongside abysmal memory capacity. Not. With. A. Ten. Foot. Pole.

 

[Bomby569]

They paid the premium beforehand,

They continue paying after, the horror stories of having to pay shipping for 2,3 or more RMA's (in the US they have to pay the shipping)