Nvidia Clock-Blocking: The narrative of what seems to have happened.

News agencies are picking up on this story, but they don't seem to be getting the story right. It is like they're spouting old wives tales in an echo chamber.

So I'm going to do this. Sorry for splitting the topic, but I want to make this easy for the news to find.

Let's start at the beginning:

A long time ago on a Nvidia drawing board far, far away....


[Picture from this post: http://forum.notebookreview.com/thr...idia-mobile-gpus.771169/page-12#post-9926255]

...a mobile GPU was designed with less VRMs than had previously been on mobile GPUs of it's class. The trip from drawing board to launch is not one of months, it's a long process. More than a year easy, probably two. Why was it done? To save on parts? To limit voltage for overclocking? A little of both? Who cares? Nvidia did it, they knew they did it, and they understand the effects of doing it.

[EDIT] Correction/Clarification [Thanks Octiceps]

So: Drawing Board... yada, yada, yada..., orders for manufacturing and assembly..., yada, yada, yada..., distribution and launch in OCTOBER, 2014.

Laptops come out. Drivers still permit overclocking mobile GPUs like the 980m. Companies lend these out to reviewers, even those with a history of overclocking as part of the review. Reviews are published or posted.

http://forum.notebookreview.com/threads/msi-gt72-dominator-pro-gtx-980m-review-by-htwingnut.763709/






Early adopters of laptops spread by word of mouth the same results about overclocking during the all important holiday sales season. Nvidia says nothing about a plan to prohibit overclocking.

Bunch of happy holiday shoppers, yada, yada, yada..., Christmas and Hanukkah, yada, yada, yada... nothing.

Then, on 2/11:



Isn't it "a little too convenient" for Nvidia to forget to tell us this, not only during the holiday shopping season, but also during the time retailers usually give to return products after the holidays?

After the misrepresentation of the details about the desktop GTX 970 to consumers, does Nvidia deserve the benefit of doubt in this case? Is "I forgot" or "Ooops" good enough a reason, or have they used up their supply of "plausible deniability" and this is now more consumer deception? These laptops had been out for four months with Nvidia's complete silence.

I myself made my purchase decision heavily factored on that +135MHz, I always added that in when looking at the specs of the GTX 980m and it affected my choice in terms of choosing better cooling models. I picked a GT72 Dominator Pro 211. I could have save over $300 if I knew extreme cooling solutions weren't going to be necessary.

I n any case, the die is cast for both sides. Yes, Nvidia has blocked overclocking. At the same time, that they left it available for a time, they allowed proof to accumulate that at least the 970m/980m can be overclocked up to the maximum allowed all other [Non-Maxwell] mobile cards made by Nvidia, which is +135MHz. The safety of overclocking the 980m & 970m has been proved and Nvidia can only offer empty words to the contrary that just don't hold up against all the overclocking results.

So I make this plea to the news outlets: Don't go on old assumptions or just repeat the corporate line from Nvidia. There is more to the story.

 

I agree with the stuff you've written there, except one thing, some of the 980M and 970M MXM cards have 4 VRMs along the top of the card, just like the MXM cards from the Kepler generation. But yes, if I had to choose a card, I'd choose one with the 4 VRMs, because I'd imagine the power supply for the 4 VRM cards would be more robust than the 3 VRM cards.

I don't like NVidia locking down their new Maxwell cards, and I also don't like how they've gone about it - you've encapsulated that well I think.

 

Yeah, I hate how so many news outlets have swallowed the Nvidia company line hook, line, and sinker. Whatever happened to journalism and objectivity, or at least presenting both sides of the story? As much as we usually hate on them, WCCFTech actually seems to be one of the few who takes our side and gets it.

Also, I think this might be a worthwhile read:

 

Thanks for the Tech Inferno link. Do you know where Kenglish got his info from (I know he's knowledgeable, but always good to have sources)? So the 980M actually has a theoretically better power delivery system than the 970M and the previous generation Kepler cards then, with the 970M having the same power delivery system as the Kepler cards. If that's the case, then we can't blame the overclocking block on the what looks like a weak power delivery system - because the power delivery system is just fine, better than previous generation!

 

Probably his own knowledge. IIRC he goes to a engineering school, plus you know he loves hardware modding/tinkering like crazy. You can ask him in that thread on T|I.

 

I have 4 VRM`s on my 970M

haha

EDIT: Those arent VRM? Damn :/

 

yada yada, who pays more money is the one who is right.

 

No my friend, if you mod one you need to mod them all for all eternity.
One flash and Bob's your uncle for ever.

 

Why not have both choices. There are pros and cons to each, more cons than pros in the case of vBIOS mod IMO. Personal preference, really.

 

I'm not really anywhere near that level of a techie and I didn't see that info before. I'm more of a "see the big picture" type with moderate understanding of tech, but nowhere near that level as it concerns electronics and I wish I did see that because it changes the big picture. On one hand, it blows up part of my narrative because the consensus is that voltage regulation is adequate in light of that information. At the same time, it makes Nvidia's stated need to restrict overclocking make even less sense in a way.

It still doesn't excuse the delay in announcing such a lockout on overclocking until four months after launch.

 

There are a lot of things that don't make sense right now. *Sigh*

 

Except that NVidia is putting the bug back in next patch.

When will they ever make a driver that has no bugs? *sigh*

Happy overclocking

Sent from my GT-I9300

 

LOL really??