News Nvidia accused of crippling board partners' designs

Considering EV bot has been removed from all future EVGA cards they have been pushing them hard it would seem.

 

Sooo...this same argument could equally have been applied to previous Nvidia gpu's, are we to believe that Nvidia have suddenly become concerned about this?

I'm of the opinion they have found a weakness in the 600series/28nm gpu's which they are trying to hide by these measures...they have been trying to prevent excessive overclocking in the drivers and now at the hardware level......somethings wrong.

Either that or it's a marketing ploy, perhaps "big kepler" will be released without being hobbled in the hope that people wanting top performance will be forced to buy the flagship rather than OCing the pants off a lower model.

 

I smell foul play at work here... first it read:

then it was

just as I was going to write a reply here...

 

Death of the custom PCB layout starts here I guess... shame, I really like the Asus and EVGA models.

 

If Nvidia set limits at the hardware level, and board parners make mods to get around this - then they should still be allowed to sell without threaterning treatment from nvidia but under conditions - no warranty from nvidia and box must say something about running above nvidia's guidlines and all this should be agreed before the retail package is developed.

 

Does it have anything to do with this?

 

I would be too scared to mess with the voltages on such an expensive gpu.

£400 paper weight anyone.

 

It's not really about Nvidia specifically, as transistors shrink Electromigration is a physical phenomenon that becomes harder to avoid. Overvoltages are simply more likely to cause an hardware failure these days, and NVidia believes that limiting voltages of on-the-shelf cards will increase the lifespan of their high-range products. They don't want their brand harmed by negative press on a bunch of products run outside the spec coming back bad.

 

Seems pretty reasonable to me, the reason makes sense and rings true from a technical stand point.

As they have evolved their designs to have a GPU boost feature, perhaps they feel that the stock voltages are now pushing the cards as hard as is safe.

 

And yet we don't see Intel who are at the even lower process size of 22nm or Amd also at 28nm on their gpu's introducing similar measures??...in fact conversely intel are encouraging overclocking with their K series?

 

Again this?

 

Actually, that's true for both AMD and Intel, and it's the K series in both cases. However, the difference may be that CPU overclockers are less likely to overvolt? I freely admit that's pure speculation on my part. As far as I've done, I've never overvolted a CPU or a GPU - given how prone I am to stupid mistakes, I figure I'm safer making myself slightly less likely to fry my everyday CPU/video card.

And it's likely the case all around, the electromigration issue. It's just that NVidia's more nervous about getting a bad rap than AMD and Intel. Aside from that, Intel likely doesn't sweat it that much. Where AMD and NVidia are on fairly level ground, I'd speculate it's much less likely that if you blow up your 3750K processor by overvolting, you're going to bail on Intel and buy a Bulldozer. For AMD, they likely don't want to unnecessarily restrict themselves when they're already struggling for marketshare in the CPU space.

Seems to be a calculated risk on NVidia's part - piss off a tiny but vocal camp that wants to overvolt while lessening the risk of a bad reputation due to parts breaking down, versus that camp switching teams and overvolting the holy hell out of AMD GPUs.

 

I suppose earlier nVidia got stuck with the bill for defective chips due to overclocking/-volting. As they correctly state electromigration is something that happens (more or less) slowly over time and it might not be the apparent cause of a defect. And as also stated, as you shrink the process the threat of electromigration gets bigger.

It therefore seems to me nVidia is simply trying to avoid having to pick up the bill for its customers' customers' OCing/OVing.

 

No wonder XFX dropped them...

 

I think they are making a mistake.

That is all.

Eventually they'll see that we don't like it and neither do hardware partners.

But to be fair to Nvidia, I don't think AMD are particularly keen on overvolting any more either.

 

As much as I don't like the idea I kind of understand Nvidia for doing it.

At the end of the day it is them who loses out come warranty time.

Sent from my GT-I9300 using Tapatalk 2

 

So does this limit the voltage adjustments i make in Afterburner? (Im assuming no from the statement that these programs keep it within its spec limit)

Or is this just limiting the soldering crowd?

 

There's a reply from nvidia that basically says that - board partners can sell over volted cards but nvidia won't warranty them - the board partner will have too. It also says there will be no discriminatory treatment of companies that choose to do this.