NVIDIA Response GTX 600 Series Voltage Control

[SirPauly]

I think the market will largely determine where this goes long term.

Indeed!

 

[Dribble]

I don't see the problem - it's like buying a car from a tuners who have turned the turbo right up and then the tuner complaining the Ford won't warranty it the same as all the bog standard cars. Of course they won't cause it's not running within the specs Ford designed the engine to do.

In complaint email seems like MSI were trying it on - they sell cards for lots of money which overvolts out of nvidia spec, then expect nvidia to deal with it when they break. Like nvidia said MSI or anyone else are quite welcome to sell overvolted cards and nvidia won't punish them for it, but if it goes wrong it's their problem.

 

[chimaxi83]

Exactly, and the fact that the AIBs are giving voltage control, then taking it away (like EVGA with EVBot), should show that their engineering knows something we don't. Can only assume that running too far out of spec will kill cards fast.

 

[Vesku]

Interesting quote. I wonder how you would feel today if you bought an EVGA GTX680 Classified for a premium price only to find out that Nvidia might not stand behind it because EVGA made it with a volt mod!:$

You don't RMA your EVGA card to NVIDIA you go through EVGA. It's not the consumers job to parse spats between companies.

 

[3DVagabond]

Everybody knew the Keplers had locked voltages before making their purchase, if you don't like it go AMD ?

These models from EVGA and MSI were sold/advertised/reviewed/ with overvolting as a feature.

Interesting quote. I wonder how you would feel today if you bought an EVGA GTX680 Classified for a premium price only to find out that Nvidia might not stand behind it because EVGA made it with a volt mod!:$

nVidia doesn't stand behind anyone's 680. It's the AIB partner who warranties the card. Ask anyone who bought a BFG card how nVidia's warranty coverage is.

 

[3DVagabond]

Like nvidia said MSI or anyone else are quite welcome to sell overvolted cards and nvidia won't punish them for it, but if it goes wrong it's their problem.

AMD doesn't warranty cards that are even O/C'd, never mind overvolted, but the AIB's still offer it and cover it. Why won't they do it for Kepler? There's something else that we aren't being told. Either there is some retribution to the board partners for doing it, or everyone knows that adding voltage to the chips will make them fail and nobody wants to take responsibility.

 

[blackened23]

Phaedrus, who works for EVGA in China had this to say regarding nvidia's explanation of money having nothing to do with the issue:

They're lying.
Trust me.

Apparently nvidia is being disingenuous with their press release to BSN -- Nvidia is doing this for revenue, and fluffs it up with a press release from Bryan. This also hints that EVGA wasn't given a "choice" in the matter, they were forced to remove EV Bot. Everything Jacob Freeman and Phaedrus have stated, points to them being completely 100% forced to remove it. (<--this last statement is speculation, yet if you read everything JacobF and other EVGA reps have stated, it seems to indicate they didn't have a choice. I'm not 100% certain)

http://www.overclock.net/t/1312524/bsn-nvidia-response-to-gtx-series-voltage-control/100

I am ticked off that they're applying a methodology that is 100% fine for a reference card, to aftermarket cards. If they want to lock down a reference card, fine. Don't lock down cards with substantially better hardware that can handle the overclocking/overvoltage. And no, the "terms" nvidia is offering AIB makers isn't a choice..... Basically, this punishes AIB makers for producing cards that are above and beyond the reference design. One can analogously state that a "reference" intel sandy bridge with a stock air cooler would be a poor candidate for overclocking -- yet when you get good cooling that opens up many more options for you in terms of overclocking. I see GPUs as being no different - if someone like MSI can produce superior components to support overclocking with a slight voltage jump, what's the issue? And don't say that nvidia gave MSI a choice, they didn't.

Thank goodness I got my 680 lightnings early in the cycle before all this nonsense happened

 

[tviceman]

Redlined at 1050? Uhh no.

Exactly. So you're original opinion / assessment is entirely false by your own admission.

What I'm referencing is the massively repeated "GTX680 Kepler was supposed to be mid range, but haha Nvidia overclocked it and beat AMD lulz" line that all the fanboys repeat every chance they get. There's been evidence to support that claim, it's been posted here, search and Google are your friend.

I don't care what fanboys have said, or what GK104's performance is relative to anything AMD has. None of that has anything to do with GK104 being "redlined" like you said. And again, the power draw and general overclockability of the chip entirely supports the opposite of your statement. Regardless of what GK104 was originally intended to be in Nvidia's lineup, regardless of it's die size, regardless of how much people want to talk it up or put it down, it's role and price point has no bearing on anything about you claiming it being red lined.

Some guys here and at ocn.net can no longer reach the original overclocked speeds their cards were hitting when they were new. I'm one of them, and this is on voltage locked cards. My DC II that I modded hit 1500 MHz, but after about a month of that at 1.225 V and sub 40C temps, anything over 1350 was unstable.

That is anecdotal user experience, not a generally wide consensus, but also not uncommon when overclocking a chip as high as it will function stably. I had a c2d e6700 that ran overclocked and overvolted for about nine months before I started getting instability in programs that were previously 100% stable. I had to reduce clock speeds. It happens to any chip when it's being pushed to it's absolute maximum functionality. And 1350mhz without being able to adjust the core voltage is a monster overclock. Had voltage adjustment been available, I'm sure your friends at ocn.net and here would still be operating at or near those same core speeds.

Anyway, Nvidia knows what they're doing by locking it down. They didn't just pull it outta their poop shooter.

Something we agree on.

 

[chimaxi83]

Exactly. So you're original opinion / assessment is entirely false by your own admission.



I don't care what fanboys have said, or what GK104's performance is relative to anything AMD has. None of that has anything to do with GK104 being "redlined" like you said. And again, the power draw and general overclockability of the chip entirely supports the opposite of your statement. Regardless of what GK104 was originally intended to be in Nvidia's lineup, regardless of it's die size, regardless of how much people want to talk it up or put it down, it's role and price point has no bearing on anything about you claiming it being red lined.



That is anecdotal user experience, not a generally wide consensus, but also not uncommon when overclocking a chip as high as it will function stably. I had a c2d e6700 that ran overclocked and overvolted for about nine months before I started getting instability in programs that were previously 100% stable. I had to reduce clock speeds. It happens to any chip when it's being pushed to it's absolute maximum functionality. And 1350mhz without being able to adjust the core voltage is a monster overclock. Had voltage adjustment been available, I'm sure your friends at ocn.net and here would still be operating at or near those same core speeds.




Something we agree on.

You seem to be bent on e-fighting about Nvidia. I really couldn't care less dude, AMD or Nvidia, I use both. I don't care to argue, I'm just here to make fun of the video card religion.

Anyway, I never said anything about "redlining at 1050". If I'm wrong, please quote me. What I said was Kepler is close to its limit as far as its speed/voltage, hence the lockdowns Nvidia has in place. Spin my statement however you want

 

[RussianSensation]

I don't care what fanboys have said, or what GK104's performance is relative to anything AMD has. None of that has anything to do with GK104 being "redlined" like you said. And again, the power draw and general overclockability of the chip entirely supports the opposite of your statement. Regardless of what GK104 was originally intended to be in Nvidia's lineup, regardless of it's die size, regardless of how much people want to talk it up or put it down, it's role and price point has no bearing on anything about you claiming it being red lined.

Why won't NV warranty any voltage increase over 1.175V then? GK104 was not red-lined in terms of frequency since there is room for overclocked after-market versions above reference clocks. However, it seems that either NV thinks GK104 silicon cannot sustain above 1.175V long-term or they are trying to reduce RMA from fear of failed overvolted Kepler chips, which says it's all about $ and not caring about the enthusiasts. The entire issue of RMA between AIBs and NV clearly highlights that NV will not warranty any cards that are overvolted beyond spec.

Sorry, but NV is not portraying the image of a company that cares about enthusiasts if it completely clamps down on voltage control after allowing AIBs to go ahead with it. Do you actually believe that MSI and EVGA would spend $$$ before launching the Lightning and EVGA Classified and launch both of those products and later go backtrack and realize "Sorry, these products are not worth the RMA costs to us." To me that sounds very naive. What likely happened is NV let them claim warranty against failed Lightning and Classified chips and then later redacted this option after other AIBs complained that MSI and EVGA has superior products. Why would both MSI and EVGA suddenly drop overvolting out of the blue after both have spent resources on developing Power Edition/Lightning and Classified cards? It doesn't add up.

I am not buying it at all that AIBs are to blame for this. Why was voltage control allowed on GTX400/500 series? Not warranting overvolting is 100% NV's doing and if RMA costs to AIBs were significant in the past, they wouldn't have had voltage control on AMD or other generation of NV cards. It doesn't add up historically unless in the past NV warrantied the overvolted Fermi chips and this generation they stopped. Clearly, NV just cares about the bottom line. They made the cheapest reference design for GTX600 series in years and by removing voltage control, going with cheap 4 VRM power phases, crackling stock fan, extremely cheap cooling heatsinks on 670s. They are either not confident in their engineered products to support overvoltage, or are greedy about every extra cent saved from RMA because they went with the cheapest possible designs this generation that had no tolerance for over-voltage. I guess you can call that "smart engineering" from NV but to me that doesn't inspire much confidence in teh quality of the components chosen for their reference cards. Even the reference 7970 can take 1.25V and not die. Sure the noise levels and power consumption are terrible on a reference 7970 but it takes voltage without even breaking a sweat. Tahiti XT 28nm works at 1.25-1.3V but NV won't warranty anything above 1.175V on their 28nm tech? Why is that?

If someone is an NV shareholder, they should be thrilled about this I suppose. As a PC gamer, this either tarnishes GTX600 series reputation for me as being flaky and unable to safely exceed voltage levels beyond specs based on NV's internal testing of this 28nm node, OR it paints NV as a $ greedy corporation that only cares about every extra cent and not pleasing the PC enthusiast gamer. Furthermore, since NV won't cover the warranty for failed after-market AIB cards, it is ruining differentiation for NV's own AIBs. The entire market for Lightning and Classified and Power Edition cards is under threat on the NV side because what's the point of a $550-650 after-market card with top-of-the-line components if can't support overvoltage without a hard mod?

Ironically, you yourself bought the MSI GTX670 PE because it had voltage control. Why else would you have paid a premium for that card over say Gigabyte Windforce 3x or purchased it over the much quieter Asus DCUII 670? Clearly voltage control is important since it allows 670s to go to 1340-1350mhz. NV dropped the ball here for the high-end enthusiast who overclocks. At the very least, they can redeem themselves by offering "K" or some other special edition voltage unlocked chips and charge a premium to cover RMA costs.

 

[f1sherman]

Phaedrus, who works for EVGA in China had this to say regarding nvidia's explanation of money having nothing to do with the issue:

There was never any such explanation to begin with.
Even so, since when is Nvidia making money suddenly some secret agenda?

Pissing off customer base in order to maximize profit <-- How come NVIDIA didn't think of this before?

 

[RussianSensation]

Pissing off customer base in order to maximize profit <-- How come NVIDIA didn't think of this before?

8800GTX, GTX280/285, GTX480 and GTX580 had a tangible lead over their competitors. NV could do anything they wanted to during those 4 generations since they had at least a 15-20% performance lead (8800GTX was way more).

This generation is different, which is why it became a bigger deal. You need voltage control on GTX680 to beat an overclocked 7970 because a 1280-1290 mhz 680 cannot beat even an 1170mhz 7970. This is the whole point why MSI Lightning was so special - it could hit 1350-1380mhz on air with overvoltage.

If GTX780 has a 15-20% lead over 8970, people probably won't care as much that it doesn't have voltage control. Voltage control is important not just for high-end but mid-range too. It'll be interesting to see if AMD keeps overvoltage and dual-BIOS switches for next generation and NV continues with locking down voltage on GTX700.

How will the majority of NV consumers react? My guess is 80% of NV consumers don't care since they don't overclock. Out of the remaining, probably a small fraction overvolt. NV still sold cards during FX5000 and GeForce 7 series and in both of those, they got beaten easily. NV could sell a real turd and it would still sell - FX5200, FX5900, 8600GT, GTX550Ti. The fact that NV removed RMA for overvolted cards shows they have estimated the loss of RMA to be greater than the loss of income from losing potential/existing fraction of customers.

When HD6950 was $250ish and unlocked into a 6970, people still bought 570s. So again, I don't think NV really cares to about losing a few enthusiast overclockers over the loss of voltage control. It seems no matter what NV does, they get away with it and their fan base almost never criticizes them for high prices, or removal of enthusiast overclocking features.

 

[SirPauly]

If nVidia continues on this thought process -- AMD may garner a competitive advantage --- garner more sales and why I am interested on how the market reacts to this.

 

[blackened23]

There was never any such explanation to begin with.
Even so, since when is Nvidia making money suddenly some secret agenda?

Pissing off customer base in order to maximize profit <-- How come NVIDIA didn't think of this before?

Do you honestly think EVGA and MSI made the lightning and classified cards without first consulting nvidia? What warrants nvidia changing course 1 month after both cards were approved and released? Both product lines were designed for and advertised with overclocking features in mind. That's why they cost 100$ more than the reference design, otherwise what's the point?

It is no secret that from the outset these cards were designed with overclockers in mind. Come on, 12 phase PWMs? Do they do this for cards without over voltage capabilities? Its not hard to read between the lines of a press release.

 

[f1sherman]

I am not entirely sure that Nvidia disassembled each and every of tens of Kepler SKUs,
and then analyzed every single spot on PCB

• With Green Light, we don&#8217;t really go out of the way to look for ways that AICs enable manual OV.

For all we know MSI and EVGA cards in question were never Green Light-ed to begin with.

• As I stated, this isn&#8217;t the core purpose of the program. Yes, you&#8217;ve seen some cases of boards getting out into the market with OV features only to have them disabled later. This is due to the fact that AICs decided later that they would prefer to have a warranty. This is simply a choice the AICs each need to make for themselves. How, or when they make this decision, is entirely up to them.

It's entirely possible that NV is a villain, but I just don't see it in this case. There has never been this much SKUs hitting the shelves right after the launch.

 

[NickelPlate]

If nVidia continues on this thought process -- AMD may garner a competitive advantage --- garner more sales and why I am interested on how the market reacts to this.

Same here. I'm poppin' some corn as we speak. Alot of conspiracy theories going on here. As I've said before the market will dictate what happens. Even though I've always bought Nvidia cards except for once I went AMD (and it wasn't a positive experience), I personally hope this causes enough backlash to give AMD some more edge. They need all they can get right now, and consumers will benefit from the competition.

 

[Rvenger]

I wonder if Nvidia's plan is to milk the AICs to give them better binned GPUs for a higher price so they can make the higher overclocked cards instead of just overvolting them. It seems that they might be double dipping which in turn is passing the expense onto us. This is just a random thought to see if it sticks but it almost makes sense if you think about it.

 

[notty22]

Same here. I'm poppin' some corn as we speak. Alot of conspiracy theories going on here. As I've said before the market will dictate what happens. Even though I've always bought Nvidia cards except for once I went AMD (and it wasn't a positive experience), I personally hope this causes enough backlash to give AMD some more edge. They need all they can get right now, and consumers will benefit from the competition.

The amount of time assume is used in endless conspiracy theories is tiring.

We can spin other debatable topics this way even more easily.

AMD clocked the 7970@ 925 mhz because thats all the majority of them are stable at.

Further proof, is a bios update with a large voltage increase for a 100mhz up-tick. 1.25volts, all we hear is indignation that AMD engineers got this wrong.

We hear short pcb's are 'cheap', 'poor quality'. That AMD uses better components, but I seem to read about more coil whine issues with AMD cards than complaints from higher selling Kepler cards. Maybe because they are happily playing games instead of defending their purchase ?
You can only assume this /sarcasm

 

[RampantAndroid]

No [duh], if you dont OC why would you care? State a more obvious statement next time please...

Pardon my french.

This is a forum full of tech enthusiasts, with the vast majority involved in OC. We are NOT the minority, you see ppls sig? OC CPU, OC GPU, custom cooled etc etc.

Edit: Had to laugh, you OC the heck out of your Ivy Bridge which is even more hypocritical.

Edit 2: The gtx460/560 series was amongst the best, the major factor: They OC excellently, and NV fans sure as hell reminded everyone of this "feature". Now that kepler isn't great for OC, suddenly OC is not a factor? Don't be so obviously biased.

Let's skip the profanity, guys
-ViRGE

How is it hypocritical? I didn't say "you guys suck because you OC stuff" - I said "it's nvidia's right to do that and I think that they made a perfectly logical decision that I'd also make in their shoes." I didn't brick it, and it's not even a very good OC. I just got to there and stopped. I might even OC my 670 at some point, but I'm not dying to...

Whine all you want about Kepler being pushed to its limit because they wanted it to compete with AMD...but the green team caused AMD to drop their prices and release the GHZ versions. To me, that sounds like a win for everyone.

 

[blastingcap]

AMD clocked the 7970@ 925 mhz because thats all the majority of them are stable at back in November 2011 or so, when they went through qualifications, and when yields on the new 28nm process weren't as good as they are in October 2012.

Fixed that for you.

 

[RampantAndroid]

Fixed that for you.

2012? Where's your flux capacitor?

 

[blastingcap]

2012? Where's your flux capacitor?

bah, fixed it

 

[notty22]

Fixed that for you.

Have not put a lot of thought in to AMD conspiracy theories, I was just using that as a example of taking known facts and twisting them to something negative.
Could the whole EVbot venture be ending because it was very low selling and not worth supporting any more ? The max voltage limit has more or less always been there with Kepler, and when the Classified and EVbot products were debated here anyone other than happy EVGA owners ridiculed it's existence on these boards. And now we have almost the opposite.

 

[Cookie Monster]

AMD doesn't warranty cards that are even O/C'd, never mind overvolted, but the AIB's still offer it and cover it. Why won't they do it for Kepler? There's something else that we aren't being told. Either there is some retribution to the board partners for doing it, or everyone knows that adding voltage to the chips will make them fail and nobody wants to take responsibility.

On the top of my head, the actual GPUs supplied by AMD could be far cheaper per order than comparable NV GPUs given nVIDIA's stubborness in this area. There could be more stock for AMD cards (due to nVIDIA products slowing down their sales) to be shifted whereas it would giving more incentive to empty some via warranties so that the cost of storage goes down? The voltage limits of the GCN silicon is pretty good i.e. the AIBs have the data to know that the cards wont just die/or rapidly degrade over time with appropriate VRM circuitry/cooling etc? Kepler on the other hand shows worse degradation over time with voltages over 1.175V?

Im thinking its all about the cost at this point in time.

 

[cmdrdredd]

I wonder if Nvidia's plan is to milk the AICs to give them better binned GPUs for a higher price so they can make the higher overclocked cards instead of just overvolting them. It seems that they might be double dipping which in turn is passing the expense onto us. This is just a random thought to see if it sticks but it almost makes sense if you think about it.

That's not a too far fetched idea really. I wouldn't be surprised if this was the case.