Nvidia GTX 1080 Gets Uncorked: Hardware Mod Guide

The guys at EVGA have of course already been hands on with the new Nvidia GTX 1080 ‘Pascal’ graphics card for some time now, and have had plenty of time to get intimate with the chip, the power design and cooling in its Founders Edition iteration. EVGA engineering master TiN in fact recently published a full ‘uncork’ guide that reveals the new 1080 card in all its naked glory with step by step instruction on how to hardware mod the card and prep it for extreme sub zero overclocking.

As ever the guide is very well put together with tons of details that will be essential to any overclocker less familiar with the darker arts of graphics card modding. Topics covered include the tools you will need, images of the card’s key components (in high-resolution photos), advice about stripping the card, uncorking the power, voltage modding, trim pot modding, PLL voltage trim pot modding and aux voltage trim pot modding.

The guide is a solid, professional approach to graphics card modding with specific focus on the GTX 1080 card. A truly valuable resource for anyone vaguely interested in modding Nvidia’s latest and greatest.

You can catch the full article on xDevs, plus a forum thread on kingpincooling.com.


30

Germany der8auer says:

Still seems to have the 1.25 Volt limit/bug tho.

United States Strong Island says:

der8auer said: Still seems to have the 1.25 Volt limit/bug tho.


man that sucks, is this really a bug or maybe just one person having trouble? Did you experience? Microcenter is supposed to have some in store tomorrow, just ordered all trimpots and have a a-power or e-power to use.

The 1.25v limit would really suck.

K404 says:

Why would it suck? :)

United States Strong Island says:

K404 said: Why would it suck? :)


well it would save me money to still use 980 ti's

Germany der8auer says:

Strong Island said: man that sucks, is this really a bug or maybe just one person having trouble? Did you experience? Microcenter is supposed to have some in store tomorrow, just ordered all trimpots and have a a-power or e-power to use.

The 1.25v limit would really suck.


No. Checked on several cards. :(

United States Strong Island says:

der8auer said: No. Checked on several cards. :(


man what a buzz kill. If anyone is smart enough to figure it out, it would be you. And if you can't then we all screwed, haha.

It seems strange TIN would go thru all the trouble of releasing guide and not know about it.

Germany der8auer says:

I don't get it either. Either he didn't test above 1.25 on load or evga simply doesn't have the problem.

United States Strong Island says:

der8auer said: I don't get it either. Either he didn't test above 1.25 on load or evga simply doesn't have the problem.


I can grab an evga card tommorrow morning at 11am est and test it if that could maybe help.

Belgium Massman says:

der8auer said: Still seems to have the 1.25 Volt limit/bug tho.


Firmware?

Brazil Rbuass says:

Yes, firmware + BIOS needed.

Brazil Rbuass says:

Excellent, as usual, TiN guide.

United States Strong Island says:

rbuass said: Yes, firmware + BIOS needed.



man that sucks, Im about to grab my card now, do you have copy of bios and firmware. Is it possible to even flash firmware myself. This is really gonna suck if this happens on all future cards, for guys like me at least. Why would they release guide and not mention that. Its one of only reasons I grabbed card.

Germany der8auer says:

Don't waste your time atm and really don't epower it. We just tried XOC BIOS and it didn't change anything.

United States Strong Island says:

der8auer said: Don't waste your time atm and really don't epower it. We just tried XOC BIOS and it didn't change anything.




ok cool man thanks a lot for the advice. I feel so lost over here.

do you think maybe doing mod for 1.25v and putting on cold would produce some nice scores or would it still be behind a 980 ti

Germany der8auer says:

Still behind 980 Ti. Even on Ln2 with 1.25 V

United States Strong Island says:

der8auer said: Still behind 980 Ti. Even on Ln2 with 1.25 V


ok cool thanks, will hold off for now.

Czech Republic buildzoid says:

What exactly is the 1.25V bug? Does it just crash above that?

Germany der8auer says:

Yea. Just a driver reset like the card is unstable.

Argentina Alan_Alberino says:

der8auer said: Still behind 980 Ti. Even on Ln2 with 1.25 V


Just curious, what clocks did you get with LN2 and 1.25v?

Germany der8auer says:

Maybe 2300-2400 depending on the GPU

Czech Republic Elkim says:

der8auer said: Yea. Just a driver reset like the card is unstable.


Maybe the leakage lock is pushed even more then on Maxwell? :F

Belgium Massman says:

It's very funny that the press is not talking about this at all. The "fastest" card sure is faster in games and on air cooling, but in terms of technology the GTX 980 Ti is still the fastest. Primarily because of the LN2 issue of 1.25V, obviously, but also because the 1080 is clock per clock slower than GTX 980 Ti. On air, [list]GTX 1080: 2560 SP at 1607 MHz (1700 Mhz boost)[/list] GTX 1080 seems about 30% faster than GTX 980 Ti with 60% higher clocks and 10% fewer shader units. To match a GTX 980 Ti at 2.2G you'd need a 1080 at speeds far beyond 3GHz. Obviously 2500 at 1.25V isn't gonna cut it :D.

Brazil Rbuass says:

der8auer said: Don't waste your time atm and really don't epower it. We just tried XOC BIOS and it didn't change anything.


Thanks for your information mate.
You always help the community.

Cheers

United States VSG says:

Massman said: It's very funny that the press is not talking about this at all. The "fastest" card sure is faster in games and on air cooling, but in terms of technology the GTX 980 Ti is still the fastest. Primarily because of the LN2 issue of 1.25V, obviously, but also because the 1080 is clock per clock slower than GTX 980 Ti.

On air,
[list]GTX 1080: 2560 SP at 1607 MHz (1700 Mhz boost)[/list]

GTX 1080 seems about 30% faster than GTX 980 Ti with 60% higher clocks and 10% fewer shader units. To match a GTX 980 Ti at 2.2G you'd need a 1080 at speeds far beyond 3GHz. Obviously 2500 at 1.25V isn't gonna cut it :D.


Expecting mass media to cover this is a waste of time. SKYMTL at HardwareCanucks said he bought a retail GRX 1080 to do a more simple liquid metal over resistor test, so when that doesn't work he will probably do a write-up. Still.. pretty bad stuff. Seems like Nvidia is giving as much as possible to stock clocks at the expense of overclocking and performance scaling with Maxwell and Pascal (basically Maxwell die shrink).

Peru MAXKING says:

Is there any bios editor that support 1080 at the moment?

Founders Edition and Custom PCB Brands seems to have stock bios with core voltage at 1.0930v (max). Not that good actually. Usually there is a difference in voltage betweeen Reference PCB and Custom ... but not really.

I have both, and I can say ...same voltage.

Also, the max core frencuency that I got was 2113 mhz @ 1.0930v with the custom PCB (MSI GAMING X). Not bad, but still not what I hoped for.

I hope Kingpins, Lightnings, Matrixs ..have a better bios that at least allows the max voltaje 1.25v. or something like 1.2x

Belgium Massman says:

Here we go ...

[IMG]https://pbs.twimg.com/media/CkXRV1_UYAApOr9.jpg[/IMG]

Germany der8auer says:

Just a great card which can do 2800 at ~1,25 volt or is there a solution for the problem?

Indonesia speed.fastest says:

Maybe you need to access NVAPI. And i read in Nvidia website about NVAPI that "GPU overclocking APIs allows apps to run apps at maximum possible clocks. Requires NDA Edition for full control of this feature." And about NVAPI Support "NVAPI is constantly and rigorously tested not only for end-users, but for internal use at NVIDIA and our closest partners." But i dont have enough knowledge about this. But maybe this is limited from NVAPI.

Belgium Massman says:

Voltage scaling and “1.25V limit� There were some rumors spreading wildly these days regarding “1.25V limitation� or whatever on modified GTX 1080 cards, which requires here few words to explain. Hardware itself is well capable of getting to that and above voltage output for GPU core, but GP104 chip itself now more sensitive to voltage, than even previous Maxwell generation. Part of it due to thinner physical process, other part due to challenges removing heat from all those tightly packed 7.2B transistors quick enough from 21% less surface area. Those overclockers who did 2200+ MHz on GTX 980 Ti’s are well aware of all things required to achieve those high clocks. Same principle applies to Pascal generation. So if you can manage to keep GPU cooled well and have good voltage delivery to it, you indeed can push higher voltages. Cards cooled by liquid nitrogen during this guide testwork were able to run 1.35-1.4V, reaching speeds over 2500 MHz. Fact that GTX 1080’s capable of reaching 2.1GHz on aircooling without any modifications confuse lot of people, making them to think that these chips can overclock well past 3GHz on liquid nitrogen cooling. But it’s still silicon, with similar architecture, so reality is bit sour. Yes, it allow to get good performance without extreme cooling, but hides the fact that LN2-cooled 980Ti is still much faster than overclocked GTX 1080 due to more shader cores and better CPC performance. This also brings and answer to the question if overvolting can help OC on aircooling or watercooling. It does not help, due to thermal, which get only worse. Higher temperature render stability and performance decrease. GPU literally overheats and cannot run high frequency anymore, even though temperature is below specified maximum temperature +94°C. Think of it as temperature to frequency dependency, all the way down from +94°C to -196°C, with slope around 100MHz every 50°C. So just like in 980/980Ti/TitanX case, over-voltage on aircooling/watercooling is not recommended, as it gains little if any performance improvement. Don’t get this message wrong, as GTX 1080/1070 are still great cards for daily gaming/content creation and VR experience. They are fast, not power hungry, moderately cool. The only catch is that overclocking them is not as fun and rewarding as it was on previous generations, even considering all tricks involved to get Maxwell clock high. As usual, any feedback and questions are appreciated. Feel free to share link to this guide, but keep links and references intact, as guide likely to be updated in future.
https://xdevs.com/guide/pascal_oc/#voltsc

Please log in or register to comment.

Leave a Reply: (BBCODE allowed: [B], [QUOTE], [I], [URL], [IMG],...)