Software Overclocking NO LONGER ALLOWED BY NVIDIA!!!

lol, thanks for these results! ill make sure to post them in the other super-thread concerning nvidia´s overclock-block. btw, how many people participated in the survey thus far?

as for overclocking and RMAing: its difficult for a survey to cover every angle of such a non-black-and-white topic. one cant just state that every single overclock leads to a component failure. same with not every RMA being due to overclocking

Let´s take a look at some distinctive and defined examples (answers regard my stance on the right to RMA):

- User A overclocks AND overvolts the **** out of his GPU and runs those clocks 24/7 --> Hell no!
- User B overclocks to the max and keeps those clocks 24/7, but does not overvolt --> still Hell no! Even though overvolts are more damaging to a GPU, the constant stress (especially on the vRAM) very likely shortens the life span of the GPU
- User C overclocks and/or overvolts, but only uses these settings for gaming, not on an everyday basis --> now were heading into the grey zone... No universal answer possible here, since it depends on the amount of overclocked / overvolted gaming that user does. same rule applies as before: if the GPU is under constant or near-constant stress, then Hell no! If the user only plays like once every few months, it shouldnt have a deteriorating effect on the hardware, especially when there is no overvolt involved and the vRAM is treated gently
- User D overclocks / overvolts to check the limits of his hardware and verifies this by benchmarks. uses undervolted / underclocked settings on an every day basis and stock clocks for gaming --> Still grey zone, but a bit more white than with user C. Depending on the intensity of the benchmarking (all day, every day? or just once in the beginning to check the performance levels and then be done and happy with it?) i´d either say hell no! or why the hell not?!

All in all, Id say: it depends! If 95% of a component´s life was spent at stock or even underclocked / undervolted settings and it suddenly gives out during web surfing, then by all means, RMA it! If, however you constantly beat the crap out of it and it fails during a stressful benchmark, then seriously? Take better care next time and go buy urself a replacement with your own money!

that´s my 2 c on the matter

PS: i perfectly understand, of course, that you guys cannot possibly check the circumstances, under which a hardware component failed. So it makes sense to put a universal regulation in place and just treat every customer accordingly, whatever that may be (as long as the customer is aware of that regulation / policy)

 

Hi jaybee83,

we appreciate your well-thought and reasonable opinion on this matter! Over 330 people have participated in the survey so far. We will include it into our official shop-newsletter in 2 weeks so we expect even more results next month!

More numbers:

How often do/did you use your GPU overclocking profile in your laptop?
All the time / most of the time: 46%
Only during experimental phase, trying to find the limits: 17%
Only for certain, very demanding games/applications: 37%

So now your question seems to be:

How many people will buy a laptop with NVIDIA GPU at a price tag over 1000€, take their time to do GPU overclocking...
and then only use this machine for web-browsing?

10%? 20%? 50%? This calls for another survey! We will call it the "I bought a huge laptop but I don't know why!?" survey.

 

Some of the time I think when people finally have the money for their dream system they are so busy with work they have no time to use it.

 

Chances are if someone was overclocking the GPU core and the card failed on stock voltage, the card was defective to begin with. If they overclock the VRAM, the artifacts could expose user error unless other cards are failing because of VRAM. There are ways to tell what made a card fail and then decide whether it is covered under warranty or not. Personally I think there should be an option to purchase an overclocking warranty like Intel offers on their K and X desktop chips. It's extremely rare to kill a card with overclocking though. These same chips go in desktop cards and have much higher clocks and voltage than a laptop could handle. There's built in protection in both the video card and the computer to make sure thermals stay in the safe range. Most customers would not even be able to supply enough power to run the voltage up alongside their power hungry i7 with these dinky power supplies anyway.

Sent from my Nexus 6 using Tapatalk

 

@mySN.de

The proper question would be: "Would you purchase a Notebook brand that allows for customized user adjustments covered by warranty over one that doesn't"?

If you would make only 30% more sales in the first place, then who cares if you have a realistic 1% more returns because of mishaps?

I don't know about global sales, but I have half a Million individual visitor on my blog accounting for over 100k downloads.
Those are the Clevo user that want the freedom to make their own choices with the hardware that they bought:

https://twitter.com/PremaMod/status/568116942579445762

 

Tom / Meaker / Ethrem / Prema: It´s funny how diverse your opinions and comments are and still i absolutely agree with all of them

As a matter of fact, I use my machine (see sig) MAINLY for work based on the usage time. That covers about 80% of total usage and includes: Excel work, image editing, Word, Browsing, Number Crunching / Mathematical calculations, etc.

The rest of the 20% (which is my freetime) is made up of, lets say, about 5-10% Browsing, another 5-10% Movies/Videos/Clips and THEN comes the 5% Benchmarking / Tweaking / Gaming

Sure, some might say this machine is a bit overkill for my needs. BUT:

1) Especially when computational-intensive parts of my work come into play, it does make a difference if i need 1-2 or 4-5 days in order to crunch the numbers I need!
2) Desktop is out of the question, too much moving back and forth, so it has to be a powerful laptop
3) I dont like buying a new machine every year, so might as well get something GOOD to last me several years (coupled with MySn´s 4 yr platinum warranty = complete package )
4) The above mentioned 5% may be the smallest, but at the same time most important / most fun of the listed parts And the more powerful the hardware, the more fun it is
5) Other people get their Ferraris, 10 MIllion-Dollar cribs, 100-Million-Dollar Yachts, so cmon! Lemme have my 2500-Dollar pimped-up Clevo, for f***´s sake

 

I'd RMA yes, but only if I suspected the damage was done by something other than the overclock.

Sent from my Nexus 5

 

Sounds reasonable to me, and on the mobile side since there's only 1 design (the reference design) it makes things simpler. Now allow me to go OT for a bit...

On the desktop side I tend to have a very different opinion. Some manufacturer design their own flagship cards (Asus -- ROG; EVGA -- Classified; MSI -- Lightning) and advertise stuff like "enhanced power delivery" and "dual/triple BIOS with LN2 mode" which basically screams "PLZ OVERCLOCK ME". And in fact EVGA even outright promotes extreme overclocking as the main allure for its Kingpin cards, and charge a hefty 35% premium for them.

So I'd say in cases where the manufacturer blatantly advertises overclocking as a selling point, and charges a hefty premium for it, I absolutely think RMAing a dead card due to overclocking needs to be allowed. After all that's what I'm buying this card and paying the premium for right? It just seems so ironically retarded for them to say "this card is designed for you extreme overclockers in mind!" and then put in a "warranty void if operated outside of designed specs" clause.

 

https://forums.geforce.com/default/.../gtx-900m-overclocking-with-347-09-347-25/17/

(see post at bottom of page)

180 done and overclocking to be re-introduced in the next driver set.

 

Soooo, actually, Nvidia do listen to their customers ! Great news !
Talking about overclocking and "grey-white-hell" zones, i will say :
Overclocking is NOT affecting lifespan significantly ! ( i don't talk about temperatures induced by it, because we have monster cooler in our laptops ).
Firstly, physicaly talking, transistors are designed to run at "up to xxxmhz", wich is NEVER used, there is always( at least) 20% margin, then, running at higher than this, will surely cause damage, as transistor will locally overheat ( no enough time to saturate gate command, so running at higher, but reasonable frequency won't do any harm (still in specification).

Second, a typical GPU chip can last wayyyyyyy more than the time it will be actually used, these chips are designed most of the time to last 10 years.....who the f**** keep his hardware that long AND overclock it ?
so let's say you hardcore overclock it, to the limit, and it divide by 2 it's life-span (super hardcore !)..... will you keep it for 5 years ? i'm doubting, a lot !, that's without taking into consideration the fact that overcloked frequency are only used while gaming, thus reducing lifespan only during gaming
Then, all the components around GPU, especially capacitor are very easy to create, and life span is really precise, if it's designed to run for 5 years, it wont last a month more, and they are always selected in case of "typical scenarios" wich don't include 24/7 heavy gaming, overclocked or not, lifespan remain the same, and it will most likely fail before your overclocked GPU.

So, overclock don't reduce lifespan in my opinion

Elipsus

 

Transistor wear is mostly caused by the insulation breaking down and electron migration. How long your chip lasts comes down the silicon lottery again.

 

The thing is that with any electronics, your'e always playing the lottery

 

Thankfully Nvidia have climbed down this time: https://forums.geforce.com/default/...king-with-347-09-347-25/post/4466443/#4466443

 

Technically most of the load comes from transistor switching. As long as you don't overvolt the wear from each switch event is about the same. However a OC'ed chip does switch a bit more frequently.

 

See http://forum.techinferno.com/genera...t-overclocking-mobile-gpus-12.html#post124474 - overclocking on newer 980m's blocked in vbios.

 

Slightly less of an issue as unlocked bios files should still be able to overclock as normal but still silly. If they set the limit to 75mhz say instead of the 135mhz that would make sense (as this does not hit the power limit so much).

 

Meaker, i don't agree, flashing a VBIOS is still something tricky that can lead to "issues non correctible with format C:". More, for the 900m we do have "unlocked" VBIOS from older devices, but when it will come on next gen, we will have to wait for VBIOS mod......
About the 75Mhz limit, it's what..... 6% overclocking (on a 980m)?, cmon, it's like you didn't overclock, leave, at least a 10% headroom........i think 135Mhz was a good spot, it's 15% overclocking (20% for older cards), allowing significant performances boost, but still non extreme power consumption/overheating

 

It's about the most you can guarantee on the 980M without power throttle anyway