[VRZ] Nvidia's Green Light Program Overclocking Limitations - Origins and Implications

It should be titled:

Nvidia fails due to arrogance.

When I was w/AW, I remember Nvidia CONSTANTLY telling us that there were no driver limitations, etc...Even when the code was disected and slammed in their face, the policy was deny deny deny.

Now, it's become standard and it's okay because AMD aren't competitors...Yet, every vendor on the market is suggesting AMD cards to enthusiasts...Go figure.

They probably just want their next gpu to have a higher percentage increase in performance compared to last generations. Not sure why they did that as it only hurts those who overclock, but if all the high end products are being bought, they are enthusiasts who try overclock anyways

That's actually a bold face lie...Especially considering they threaten/threatened vendors that "failed" to adhere to the new standard.

They also removed the EVbot without actually telling EVGA...The customers were the ones to alert EVGA originally to the problem...That's kind of sad that, Nvidia can't even make a phone call to tell EVGA that they virtually just rendered an entire line of cards, worthless.

Regardless how Nvidia wants to present this, there are a few things that don't add up correctly.

The GTX 500 series allowed overvolting through software and through things like the EVbot plug. (FACT)
The GTX 500 series could be overclocked and overvolted to the nines and yet there weren't excessive failures among that lineup. (FACT)
The GTX 600 series cannot be overvolted and the overclocking is there, but not game changing. (FACT)
Something must have changed in the card's construction OR Nvidia's mindset in order for this change to have been made. (ASSUMPTION)

Things that could have changed:
1. Kepler is more sensitive to voltage changes. Nvidia may know something we don't about voltages. More volts over a few months may damage Kepler
. 1a. Higher voltages may cause more premature card failures
. 1b. This would cause more RMA's and thus more expense for Nvidia
2. TSMC's 28nm process may be having yield issues. In other words, Nvidia's Kepler's rolling off that assembly line are just barely meeting QC standards at stock volts, so Nvidia knows that pushing them in the power area might blow them up.
3. Nothing has changed at all as far as chip durability or TSMC's yield quality. This would suggest three things as to why Nvidia has locked the voltage
. 3a. Nvidia wants to reduce the number of failed chips. Fermi's were known to handle a decent voltage bump if they could be kept cool enough, but their failue rate was, from what I've been able to find, quite low. If this has not changed, Nvidia wants to save some $$$ in RMA dollars.
. 3b. Nvidia wants to make the next generation chips a really big improvement. An overclocked and overvolted GTX 480 could hang with a GTX 580 at stock, which gave many GTX 400 series users no reason to upgrade and thus less $$$ for Nvidia. By locking down the overclocking potential, we aren't able to stretch the current generation as much.
. 3c. A combination of both. Nvidia wants to make more money on the front end by not having to replace as many cards and also making it more enticing to upgrade when a new generation comes out.


If #1 or #2 is true, then Nvidia gets a pass, but they should explain this to the consumer. We really would like to not hate Nvidia, but we need to know this sort of thing to MAKE that determination.

If it's #3, (as I suspect) then Nvidia should rightly so be condemned for squeezing yet more cash out of the consumer. They make the product so they ARE allowed to do what they want with it, but we get to vote with our wallets and may choose to give business to a competitor as a result.

That's my story and I'm sticking to it.

Not entirely accurate, FYI.

The 590's had excessive failures due to manufacturing errors (revision 1) which were later "fixed" but, by then the cards were completely driver locked...

The 580's were voltage locked as well but, only on stock versions, on the SC/SSC/FTW/Classified etc (same with MSI/Asus etc) there were scaling locks but, the OCP still existed.

Any 590 user will tell you that the locks were in place immediately following the release of the card...And in fact, as I stated earlier, Nvidia's stance for 6+ months was 'deny deny deny'...Only in Sept/Oct did they actually come out and admit they were voltage locking the line.

So, this practice has actually been in effect since the 5-series...Arguably early Fermi.

Source? As far as I know, EVGA removed it because it would invalidate the warranty. Evga could keep it on if they wanted to swallow the cost of the RMAs

Read the EVGA forums, much?

http://www.evga.com/forums/tm.aspx?m=1755509

A CUSTOMER had to inform them that Nvidia removed the plug so, your information is entirely incorrect.

Only after Jacob + Several other employees of EVGA contacted Nvidia, did they get an official answer. (Several hours later)

Again, #Nvidiafail

This type of thing is why I've always disliked nV. A company that uses underhanded tactics to edge out against it's competition tends to eventually use those types of tactics to take advantage of their own customers.
There's absolutely no excuse for taking away voltage control while you have customers who paid premiums for cards that were made for voltage tweaking. The decision to force their sudden new standard for reliability in spite of knowing this is just disrespectful. It would be bad enough to make this choice for 7xx and on, but to remove features on cards they've already sold is horrendous. How can people buy nV cards with confidence when they know something like this can happen anytime?
I was strongly considering trying nV again until this happened.

Ammm, could you please clarify how NVIDIA removed the plug? Are you saying that AIC's send their desigs to NVIDIA which then, ships everything to the AIC? I thought that NVIDIA licensed the designs and sold the die's but the rest of the process was handled by different companies?