Very bad idea. They won't bother testing and finding the optimal OC for each and every GPU they are shipping laptops with.
Hell, Asus has done that with their GTX 670 Top series cards, lots of cards failed to work reliably at "factory" clocks, they ended up issuing a bios later which decreased the clocks some. Should be up to the user, OC or not, imo.
Playing devil's advocate here, but shouldn't the OEM then be shipping the laptop with an existing overclock already on it? Or offering custom BIOSes to flash on to chip? One that is the base line for running nice and cool, and one that is for when you want to push the card?
The alternative I see is coding the BIOS to include a "maximum GPU clock" parameter, and then having the driver read that field.
I'm not defending anything here, I'm merely suggesting that OEMs probably view this as a good thing (and probably help drive for this) as as I understand things, if you OC through the driver and kill the chip there's no way to tell that the chip died was overclocked. If you flash the BIOS, at least then you can look at what's in the BIOS.
Replacing chips the user killed by OCing them isn't something the laptop manufacturer should be forced to do.
The OEMs can market their notebooks with "OC enabled" if they designed a beefy heat dissipation that allows for safety. Otherwise if its a slim mainstream notebook, it would be off by default.
By going with this approach, NV is not responsible for the damage since OEMs will have full control. So there's actually no reason to block OCs on the side of "potential damage or warranty issues", since NV wouldn't be the one to bear the brunt of RMAs.
In fact, in the current situation, notebooks just need to mention "OC voids warranty" and that's it. Nothing needs to be done at all. Users OC at their own risk.
If this stays as is, I'm never buying another Nvidia product - be it Tegra, desktop GTX, or notebook GTX.
Beyond absolutely ridiculous.
The OEMs can market their notebooks with "OC enabled" if they designed a beefy heat dissipation that allows for safety. Otherwise if its a slim mainstream notebook, it would be off by default.
By going with this approach, NV is not responsible for the damage since OEMs will have full control. So there's actually no reason to block OCs on the side of "potential damage or warranty issues", since NV wouldn't be the one to bear the brunt of RMAs.
In fact, in the current situation, notebooks just need to mention "OC voids warranty" and that's it. Nothing needs to be done at all. Users OC at their own risk.
The same thing happened in the old days with CPU overclocking, via FSB. You can't stop stupid by punishing all the legit users.
IMO though OEMs asked Nvidia for this is what I think.
Well, such chips exist. They're called EEPROM which are a special kind of Read Only Memory - ones that can be written once. It shouldn't be impossible to at such a chip to the circuitry as they are quite common in things like that. In fact, they even usually use a pretty "low tech" method to ensure that they are not hacked - they black out the part number so that you can't tell it apart from the millions of other 8 legged chips (well, 8 IIRC. It's been a while since I dealt with those).But then I circle back to asking what happens when the chip fails from a bad OC - who should be paying for that? From nvidia's perspective, it's simply easier to not allow OCing than try to solve this problem.
If they had a byte that could not be cleared or decremented that listed the highest frequency the GPU had run at was (and an easy tool to view that on a dead GPU) then I could see the argument for allowing the user to do stupid things...since THEY would be paying for it.
It's blocked via the driver. I'm assuming there is nothing stopping you from modding the driver or flashing a custom vBIOS.
Besides most laptops are limited by their power brick and cooling TDP anyway.
IMO this is a smart move on NVIDIA's part, it stops 14 year old's from releasing the magic smoke from MOSFET's and boiling Caps.
I really didn't realize people bothered to OC a notebook GPU - for the same reason nvidia just killed it: limited heat dissipation abilities...
The vendor should be doing the OC, NOT the user. Because if someone burns out their chip, odds are they're going to RMA it for repair, and the vendor will be paying for the user's poor decisions. Don't you think this move by NV might be to benefit the OEMs?
What it allows is for them to charge you more for O/C'd models, and that's it. Anyone who honestly believes anything else is naive.
Last time I did that was in 2005 when I burned up my AMD Athlon FX 64 3500+ @ 3ghz on the stock heatsink... My friend said to me ya that's why you need an aftermarket cooler when OCing LOL. Leason learned the hard way.I can't remember the last time I read a thread where someone burnt a component O/C'ing.
The only one that is naive is nvidia for thinking they can go around ripping people off and lying to loyal customers.
Unfortunately their customers are paying for their screw ups. Hopefully people will learn from all this and move forward to keep nvidia in check.Nvidia knows exactly what they are doing
UH no all GPUs can be OCed. Also the 660ti was a fail card with cripple memory bus.Regular cards you can't overclock, and Ti variants can be overclocked