NVIDIA finally get an official response out to the RTX 30 series “debacle”

It was all going so well…

EVGA GeForce RTX 3080

I remember on the morning of the RTX 3800 launch I tweeted that it actually felt like a big thing – like a console launch. The PC world was excited for the step-change NVIDIA was about to introduce with the 3080 cards. The Youtubers had been gushing over the review models and I’ve seen more arty camera angles in videos for these GPUs than you get in a pretentious European arthouse movie.

Then the wheels came off. First up was the fact that, yeah, well you couldn’t actually get hold of one.

A “paper launch” it was described as by various PC luminaries eager to get one over on the big boys. But then, something weirder happened. A few people started to get hold of some of the third party cards, ourselves included and that’s where things went from a botched launch for NVIDIA to a full-scale PR disaster.

The thing is when you launch a flagship product the last thing you need is for there to be something wrong with it. Unlucky. NVIDIA isn’t the first to fall foul of this.

Iphones bent when you sat down, Samsung phones exploded in your face so maybe the occasional Crash to Desktop could be overlooked. But this is the PC fraternity here and they smelled blood.

People were shelling out a large amount of cash – often over the odds to get hold of this new game-changer and maybe, just maybe, there was a tiny design flaw that was actually a massive issue.

Well, the same Youtubers who waxed lyrical before the launch, mainly because they were given preview hardware and didn’t want to bite the hand that fed them have now turned the lens on the manufacturers and have magnified the problem enough so Nvidia, almost a fortnight after launch have finally felt compelled to come out and speak about what’s going on.

Cheaping out?

The issue is suspected to be around third-party manufacturers swapping out some of the capacitors close to the GPU for, and here’s the knife, cheaper ones. These cheaper components seem to introduce noise into the circuit and when the GPU is clocked to high levels, yeah well, hello desktop, wasn’t expecting to see you so soon.

Now when you are paying top dollar for a piece of hardware you can quickly get frustrated if you think a manufacturer has cheaped out for a few cents and ended up making your card worthless.

Of course, things aren’t that dramatic – they haven’t exploded on an airplane (hello? Is that Samsung’s marketing department?) but the PR optics still aren’t great.

The crux of it all is that NVIDIA gives third parties the reference specs to build the cards but within that, there is a degree of component flexibility.

NVIDIA in its statement says that they work closely with the manufacturers of their partner cards that are based on custom designs but also go on to say that the capacitor groupings are not indicative of quality – seeming to suggest that nobody has cheaped out here.

They told PCWorld’s Brad Chacos, “Regarding partner board designs, our partners regularly customize their designs and we work closely with them in the process. The appropriate number of SP-CAP vs. MLCC groupings can vary depending on the design and is not necessarily indicative of quality.”

It might just be me but it’s taken nearly two weeks for that so I’m slightly underwhelmed.

Either way, it seems that’s dialing back the clock speed can reduce the issue pretty much out of existence and there are some new NVIDIA drivers that also seem to address the issue (presumably by doing just this) but the internet may well not be sated by this.

In truth, it’s an issue that a lot of users may never encounter if they don’t push their 3080s to the limits, but by the very nature of the people who want their hands on this stuff first, that’s exactly what they set out to do.

Regardless of four lines from NVIDIA though, it’s safe to assume that there’s been a lot more than four lines worth of dialogue over this between the GPU giant and it’s partners, especially with AMD’s new cards on the horizon at the end of next month.