Gigabyte releases statement regarding SP-CAP/MLCC Capacitors on GeForce RTX 3080 GPUs

[WYP]

But isn't the whole issue that a decent number of cards are having issues without being overclocked? It's happening on some cards when they reach around 2GHz which GPU Boost will reach on some cards in default state. I don't think a anyone considers GPU Boost to be overclocking, it's what the card does by itself without any user intervention.

The whole capacitor issue is due to bad Nvidia drivers and a spikey GPU boost profile. The drivers solve the instability.

When I said that these cards are not designed for overclocking, I am saying that's why these GPUs do not need the best possible caps configuration. The current setup is overkill as it is, now that Nvidia has fixed their drivers.

The reason for the instability is because Nvidia had a faulty profile in their drivers which caused some cards in some games to spike too high in GPU boost and take all the energy from the caps before the power system can react and compensate for the higher loads. This, in turn, cause games/software to crash. This is also why cards with worse caps had more issues.

This is why driver have fixed the issue, and AIBs are not recalling their products. AIBs have done nothing wrong, and now engineers have to justify design choices which aren't even a problem.

 

[AlienALX]

All the driver does is fix GPU Boost. Less large spikes, lower load on capacitors and increased stability.

It's not the clocks that are the problem, it's the spikes. The spikes drain the power from the capacitors and caused instability. Fixing GPU boost in drivers fixed the issue.

Clock spikes caused the capacitors to drain before the main power circuitry could compensate. This was an Nvidia problem, not an AIB problem.

Just watched most of Steve's downhill video (lol) and he reckons that the cards were boosting too high and crashing. Mostly because of the drivers. Apparently board partners got them after the press.

Until that stage they use Furmark to burn in and test cards, but apparently it doesn't boost as high (or high enough to show the issue). Once in games they boost higher, hence the crashes.

So apparently he thinks it's a failure with Nvidia's black box, which tests the cards before they are packaged and sold.

And it all kinda smacks of cards simply boosting higher than they are capable of.

 

[NeverBackDown]

I'm just sitting here finding all this news about bad Nvidia ironic. People been saying for past couple years how awful AMD are.

I'm sure people will forget about this in a month and go back to thinking Nvidia are perfect.

Meh I'm sure the people with all the bots who bought the cards are still happy lol

 

[Warchild]

I'm just sitting here finding all this news about bad Nvidia ironic. People been saying for past couple years how awful AMD are.

I'm sure people will forget about this in a month and go back to thinking Nvidia are perfect.

Meh I'm sure the people with all the bots who bought the cards are still happy lol

Sadly the bots have probably loaded them into their mining farms by now. Wonder how this instability behaves when mining

 

[Warchild]

lol check out Inno3D response

Quite bold...