Software Overclocking NO LONGER ALLOWED BY NVIDIA!!!

http://forum.notebookreview.com/threads/no-more-overclocking-on-nvidia-mobile-gpus.771169/

This sucks. Big time. I know we can mod vBIOS, but it will take so much more effort by constantly flashing the vBIOS to tune appropriately. Grrr. I hate you Nvidia!

 

Guess who's not updating their drivers? This guy!!!

 

I dont care, im still using stock vbios (with throttling) on that gtx880m and it runs cool & quiet. Thats all I care about.

 

Vbios overcloking then .

HTwingnut : You should take into consideration that there are A LOT of noobs RMA'ing their laptops because they play with NVinspector/Nibitor after seeing one youtube video..In France for instance the specialised forums are FILLED with : OMAGAD i have a black screen, i just flashed/overvolted XXXX like this video explained it to me....oh well i'll just RMA it they'll never know . The next week they get a new laptop from their (great) reseller...

So i can understand Nvidia, they might have a lot of return because of stupid and unconscious noobish users.

In another way they might want to "casualise" the notebook market in order for a gamer to change his laptop more frequently, that indeed sucks.

 

1) choice is always good
2) nvidia is behaving irresponsibly, calling a function that has been there for over a decade a "bug" thats been fixed now
3) everyone is responsible for their own actions and should act accordingly and in an appropriate manner
4) i dont need anyone telling me what to do with my hardware

Sent from my Nexus 5 using Tapatalk

 

Sad thing is, the way we are currently headed, manufacturers and developers (like Microsoft and Apple) will continue restricting user tweaks and options.
The less people fiddle with their stuff, the lesser customer support they have to provide.
Advanced users may have to pay extra for "unlocked" versions of stuff (like Intel is doing with their "Extreme" processors).

Imagine nvidia coming up with a whole new tier level of cards, just for "experts", which have the option to be OC'd (software-wise ofc).

P.S. (fixed)

 

Nibitor is one thing completely different altogether. You are changing the vBIOS settings. nvidia is removing support from drivers and from the cards themselves. The +135MHz was quite limiting but at least it was something, and vRAM usually has a lot of extra head room as well.

 

That is a good point about Intel CPU's, I didnt even think about those. They have been doing their "K" CPU's since the i7-2500/2600 series.

 

id be happy with a K-series equivalent GPU class series, just please dont lock it ALL down! intel even offers a protection plan for overclockers, so why not use that kinda thing in the mobile gpu sector as well?

 

I know but that doesn't change the fact that some people enjoy tweaking their GPU when they don't know sh#t about it . Nibitor or NVinspector.

 

The problem with that argument is that it equally applies to desktop users (and perhaps even more so). I can't count the number of times I"ve seen posts like "HALP I flashed XYZ's super ultra mega hyper awesome custom bios that let's you overclock OVER 9000 but now my screen won't turn on!!!!" on various forums, or people who think they must max out their voltage tweak in Afterburner before adjusting clocks.

 

You can't compare the two.

Screw up vBIOS settings or flash, you can't boot into Windows and end up with an expensive paperweight.

You mess up with software overclocking? Worst thing that can happen the display driver crashes and recovers, PC doesn't even shutdown or reboot and life goes on.

 

Yeah I've seen way too much of that. People get cards and flash an extreme bios mod and fry their cards despite the million warnings they had to get through to even get the BIOS in the first place. Mobile users are smarter than that generally.

The fire is just starting...

There is something that could make overclocking dangerous on the 980M. It has 3 VRMs across the top, not 4.

Here's the 970M with the normal VRMs.

http://rjtech.com/shop/images/detailed/970m-f.jpg

Now compare with the 980M

http://rjtech.com/shop/images/detailed/DSC01680.jpg

I don't know the significance of the extra inductor but the missing VRM is worrying.

Sent from my Nexus 6 using Tapatalk

 

I disagree, overclocking and reaching the low 100°C (no secure shutdown...yet) and your card gets its life expectancy cut in five...

 

How do you reach 100C when throttle temp is ~90C?

And I wasn't referring to temps at all...

 

Maybe lv2go folks will find a way to remove this bug NVidia has purposefully introduced to our drivers.

NVidia should make the x70m and x80m OC editions.

As for life expectancy, never had issues on any laptop that I had OC and it ran hot. I don't factor in reducing life expectancy ever, never read of someone hardware failing prematurely because of OC, only if user bricked it. The life expectancy crud, I categorize it as a myth. Intel, AMD and NVidia have made mobile hardware to run at 105c since I've been buying DTR 15 years ago.

 

They will. They'll sell them next generation as GTX 1080M and 1070M.

 

Big difference. Those are usually OC editions and companies like EVGA provide lifetime warranties and will replace bricked GPUs even though its not official warranty. And they are sold either water cooling or beefed up cooling. Big difference from Clevo who only provides warranty for 1 year stock and a very very limited warranty.

 

Sure once you hard mod or soft mod your GPU, that's at the user's own risk. But it shouldn't be restricted. Funny thing is users can still overclock with nibitor or a vBIOS flash, so that's a moot point and it was never officially supported by warranty to begin with. It's the fact that Nvidiot ins't supporting even modest software overclocks out of the box.

Nvidia only supply's the GPU chip. Alienware, Clevo, Asus, and MSI manufacture their own boards. Unless there was a clear failure of the GPU chip, I'm sure the OEM's eat a bulk of the warranty related to those failures, not Nvidia.

 

wrong, clevo provides two years of warranty as long as you buy the gpu inside a machine. you cant compare buying single mxm cards with getting a run-of-the-mill desktop card! as for mxm cards, one year of warranty is actually pretty good, the usual u get otherwise is DOA 30 days, max!

 

My laptop has 1 year warranty. Not two. There isn't a disclosure I've seen that says GPU has two years.

 

One year as standard, obviously replacement parts that you get in that warranty are still covered for the term of the warranty, there are extension options if you would like it longer.

 

Is there a valid reason as to why NVIDIA prohibits overclocking? does it shorten GPU life?

 

Yes overclocking will always shorten the life of a chip unless partnered with an undervolt of a similar magnitude.

 

I was simply responding to wickette's comment that "...some people enjoy tweaking their GPU when they don't know sh#t about it" and saying it applies equally to desktop users, since it's not a laptop exclusive phenomenon.

Also EVGA is just one company, and them replacing bricked GPUs for free doesn't mean other companies do the same. And it still doesn't refute my point that desktop users are equally (if not more so) prone to doing stupid sh!t they shouldn't be doing.

 

Modding the security throttle OFFSET
Throttling doesn't mean not reaching the 100°C
Only the GPU die is throttle regulated (thermal control process), not the Vram ships (older nvidia models) not all the other components on the GPU.

Don't tell me that no one fried its Nvidia GPU because of an OC, please don't ^^ ! We all know that Overclocking = increased risk for our component IF we don't know how to manage it.

Jaybee for instance tweaked heavily his 4790K in his batman Clevo why ? because he sure knows how to do it and he knows how a tweaking can be safe in the short/long term. Someone else who reaches the 98°C bar but doesn't see any shutdown or doesn't even monitor his thermal throttling we'll keep it like that...for some months before RMA

 

delete double post, thx

 

The VRAM chips don't run anywhere near as hot as the GPU core, that's why they're fine with just thermal pads.

Never fried an Nvidia GPU due to OC/OV but have set fire to an AMD card before.

 

At the risk of catching flack for saying this..... I've NEVER had a failure due to over locking. My last laptop was an old Lenovo w700 with the CPU over volted and over clocked qx9300 at 3.5 Sucker ran in the high 90s. My desktop is 6 year old, same gig. Over clocked and over volted running in the 90s, still do this day. The desktop sees heavy use for video editing. We could go into theory of electron tunneling and whatnot but you run into the theory vs. Practice debate. In practice I've yet to have a failure

 

Same here, my BNB Toshiba which is almost 4 years old now still runs fine even though its crappy cooling system meant both the CPU (740QM) and GPU (350M) went over 80C when gaming. And yes I had a 10% OC on the GPU.

 

good thing to live in the EU then standard warranty two years by law

hm lets see.... thus far ive had:

1 mobo / 1 psu / 1 hdd / 2 gpus / 1 display controller / 2 ram sticks and one gpu watercooling block crap out on me

but all this was in a total time span of 18 years, so yeah...

Sent from my Nexus 5 using Tapatalk

 

That's not the issue. These OEMs market their GPUs as over clock ready, MSI, ASUS, EVGA, they all do it. If Nvidia introduced no OC bug to desktop, they would have issues with their business partners.

Notebooks are opposite, they don't was t people over clocking. Wouldn't surprise me if Asus and others asked Nvidia to prevent OCing. Wouldn't surprise me if this one s a problem with the notebook OEMs not Nvidia.

 

Clevo doesn't provide any warranty to end users, your reseller does.

 

Not really, the AIB partners could simply say theirs is a "certified factory OC that's 100% guaranteed safe". I mean they give you the same warranty whether you buy the reference or factory OC'd cards, so in a way that OC has been "pre-approved". nVidia could still introduce the OC bug to disable end-user OC citing the exact same excuses.

 

I could talk about Silicon lottery and the constantly increased security and quality of the components over the year but wether you like it or not : OVERVOLTING/OVERCLOKING BEYOND a safe point reduces the life expectancy of your product, if the manufacturer predicted that your chances of having a HW failure before the warranty expires is near zero, with overvolting that might not be the case anymore.


I'm not saying that Overclocking reduces the life expectancy of your product, I'm saying that dumb overcloking, (no check of temps etc) does. You can safely overclock your card/CPU, but only people who knows what is thermal throttling, whzat is stress test can know ! You won't find a lot of those people around a forum like NBR, but A LOT of people "play" with their product just by googling : Increase performance for COD.

Why Xotic talks about SAFE OC ? Because it's stable, and speaking of temperatures its under the throttlythrottle bar.

 

Now that I agree on.

 

I've seen chips rapidly age due to the overclock, they lasted 5 years instead of about the expected 6-7.

I've not directly killed anything either.

 

Whether it has a significant impact on its usable lifespan is a different question though

Edit: See you've already addressed it in your post above

 

not quite correct, there are a lot of resellers out there who send in parts / get their parts directly from clevo, in that case the reseller is just a "middle man". in one example (CEG), youre actually dealing directly with clevo service centers, thus your warranty runs directly through clevo

 

You... really have no idea how Kepler/Maxwell work... do you? Once you hit the thermal barrier (92c for Kepler; unsure for Maxwell, but it is at a similar point; probably 90c) your card downclocks. And that thermal barrier is un-change-able. Even with the modified vBIOS we're so fond of? It is not change-able, at all. It just isn't. And if you go and make your own custom vBIOS and relax that limit further past their TJMax? That's just your own damn fault. These cards don't hit NEAR their TJMax. Never. Not once. You can try, it will not happen. Go right ahead.

 

My card can reach frequently the 92°C then go above slightly, down slightly you don't seem to know a thing about thermal stress on materials :

Put an ugly glass near 100°C, cool it fastly, re-heat it, it will break. That's the same thing for all materials, if you go high temps-low temps the stress the material receive is above a stress he'll receive in a constant way, it's the same principle, it's automatism, thermal regulation, you reach 93°C, you go beyond then above frequently for six months...a sinusoidal plot.

Wether you like it or not the higher the temps are the higher the failure rate....

The throttle offset i was talking about was about the intel CPU in the P35X V3 for instance.

And to close this subject : Temperatures aren't all, you can overvolt a capacitor for instance if he doesn't support it, he'll make sure you'll know, temps or not

 

Remember that your warranty is always with the company that you purchased the product from. Clevo Service Centers may be used by some companies to provide repair/warranty services, but the warranty cover is still the legal responsibility of the seller and not with Clevo or an official service center.

 

1 - The TJMax of the chips are 100 degrees, so it doesn't hit near the thermal limit even if you DO go slightly over 92c.
2 - I know about thermal stress. I've busted thermometers in the past by alternating putting them in boiling water and ice water. These chips don't cool that fast though. You'd need to heat up with furmark with an AC vent blasting on them and max fans then close furmark, let it instantly cool, then repeat the process to see some of that thermal stressing here.
3 - We are not talking about CPUs, so this point is invalid.
4 - The max voltage on these cards is 1.1v (at least on 780Ms etc) with modded vBIOS. With stock, voltage adjustments are not possible. So overvolting, as far as nVidia spec, is impossible.

You have no arguement.

 

If your 980M reachs 92C, something is wrong.. You need to get it repasted... PC Specialist uses MX-4 paste which is junk and kind off explains this.. Even with a heavy 30% core + memory OC, I reach 80C tops with IC Diamond paste...

 

NONO My card barely reaches the 76°C I was just talking about a hypothetic scenario where my card reached the Nvidia hardcoded throttle bar 92-93°C .
I repasted my card with IC7, 62°C @ BF4 I'm really really happy with it


The TJMax of the chips are 100 degrees, so it doesn't hit near the thermal limit even if you DO go slightly over 92c.

-The TJmax temperature is the temperature limit before a auto-shutdown. It is NOT the temperature limit before you harm in the long-term or mid term your card !


- I disagree, thermal stressing is a serious problem that happens in a lot of computers, you can have a delta of 20-30° C within seconds in laptops, especially "normal" single fan laptops. Not everyone will think about putting FN+F1 ^^ don't forget that NVidia doesn't sell 900M to only CLEVO laptops with a FULL fan function.

 

I hope everyone that posts in the Clevo/Sager forum has signed this petition....

EVERYONE.

 

You mean fn+1. Fn+f1 turns off touchpad.

 

You mean like this?



SCNR

 

now THATS what Im talking about...

btw, any preliminary results from your "overclocking in notebooks" survey?

 

Yes. 13% of the people who took the survey dropped out after the first question by answering "No, I'm not interested in such kind of laptops".

Fun fact:
- 35% of the people who did GPU overclocking (or who plan to do it) answered "Yes" when asked whether they would send their dead/broken GPU in for warranty/RMA.
- The other 65% said "No" - at first. When asked a second time like "Seriously? Are you really sure? Would you not send it in?" about one third flip-flopped to "Well yes, honestly speaking, I would send it in!"

Even when you filter out (remove) those participants who answered that they only experimented with OC'ing without really using it for games and applications, the above numbers only change by 1-2%.

So bottom line: over 50% of you would use RMA/warranty for overclocked cards. Discuss!