[Release] 980M Single Undervolted vBIOS

I have decided to release my own 980M undervolted vBIOS, based off the original Alienware 980M vBIOS. I have thermal protection set to 90 degrees, forced stock boost clocks, and voltage locked to a maximum of 0.9810V. I have bundled it with Prema's nvflash and a batch file to flash the vBIOS (modme.bat). This is NOT Prema's vBIOS, it is my own, therefore it is probably not of the same quality and is subject to further revisions and improvements. This is for 980M SINGLE SETUP ONLY in the M18x R2. It has serious throttle issues in the Alienware 18. I have not figured out the throttle bug with SLI setups, but it does not appear to be an issue with single 980Ms.

SLI testing pending in the M18x R2. Will try to figure out wtf is the problem with the Alienware 18 *sigh* another demerit against the Alienware 18.....why do I even own one still is beyond me.

Flash at your own risk. Download zip attached.

Chime in with your results.

 

In M18X-R2 BIOS changing IGFX to PEG completely eliminates the throttling even using a low level CPU overclock(41,41,41,41) and only one PSU, so I think in this case there is something in system BIOS configuration causing it to throttle and a proper update could been fix it. Its a fact that Clevo machines that comes with a pair of 980M dont throttle at all so I think this inherent behavior comes from the notebook motherboard not the 980M itself.

 

There is some slight throttle I couldn't remove like Prema does with his vBIOS, not really sure how to tackle it. It appears for 1-3 seconds, then manages to resolve itself in SLI mode.....not really sure what to think. You can try to flash to your second 980M (change batch file to show nvflash -i 1 -6 980m_UV.rom)

 

You think this could work for AW17 without problems or throttling?

 

Probably not since it shares the same platform as Alienware 18, but you're welcome to try and revert back to prema vBIOS if it doesn't work

 

Any chance of bricking or mere throttling ?

 

Shouldn't brick, I flashed on my M18x R2 and Alienware 18. Worst case is throttle, but you can revert to Prema vBIOS in that case. Just use standard flashing procedure.

 

What is the ASIC if the GPU this vbios was tested on?
The voltage is contant at 0.981 through all the pstates?

My SLI has driver crashes when switching voltage.

 

Working pretty good, temps are pretty good but power throttles at 99.99 Watts. That's why you see the 1-3 secs throttle when the cards touches that power limit. If you can remove the power cap limit this would be my favorite vBios.

 

I will see what I can do, stay tuned

 

Can't wait ... I checked again my daily driver the SVL7 vBios and it goes way past 100W @1.000V voltage with zero power throttle, so I am 100% sure that power draw cap is responsible for the throttling.

But interestingly enough, my stock vBios also has that same ugly power draw cap and it throttles at 100W too. Why do people throttle a 980m at 100W is beyond me.

 

Wow, Will try this and Run Crysis 3 with the next version since the game pulls post 100W in most cases I would be throttling the whole game. +rep man for the good work.

I guess the Nvidia's failure at keeping the TDP in check unlike the 780M might have got this vBIOS shenanigans xD

 

Please note, it is for the M18x R2 only, I couldn't get it to work right on the Alienware 18, so it will not probably work well with the Alienware 17 either. I'm still trying to work out bugs, I'm working on zero knowledge so you guys are like my beta testers

 

Actually tried it for an extended period yesterday on seriously heavy games, working as well as my stock vBios with much less temperature so honestly so far super good

 

It sadness me to see how difficult it had become to tune your own hardware now compared to how easy it was in the days of old GTX7xxx and intel C2D.

 

I tested it on my SLI. It seems the 'driver has crashed and recoverd' is gone since voltage switching is between these two smal values.

I also run Fallout 4 for several hours and I noticed a 'weird' throttling.

GPU0 dropped to 5% utilization and GPU1 was a constant 100% when I had the weapon drawn. When holstered both cards were at 55 - 65 % rate.

Could be the game though....

Temps were amazing. Never got over 70 even with lots of enemies around and stock fan tables!

Keep up the good work!!!

 

As I said in the initial post, there is a strange bug in SLI systems I haven't figured out, causing my system to crash the driver every 10 seconds or so when using this vBIOS, probably power related, but even that I am unable to figure out. That's why for the most part on single 980M, the throttling is miniscule if at all present under gaming (it WILL show up in benchmarks, 100% guaranteed). I played ARK for several hours with a couple times crashing, so I think 0.9810V is too small of a voltage, will bump up to 0.9870V and if that is also too low, I will raise it only back to 1.000V. I've tried adjusting the power limits but it doesn't seem to correlate well. I actually managed to make the power problem disppear, and GPU-Z shows full clock no throttle, but my FPS went to garbage....1/5 of what it is actually supposed to be. Still playing with this.


If there is any benefit to it though, it is reduced temperature and voltage, hopefully extending the life of the 980M, as they like to kill themselves every now and then. My card ASIC values range from 69 up to 74, though I did have one at 65 that was absolute and total trash.

 

Interesting enough I noticed that yesterday when I was heavily stressing my 980m using your vBios, the voltage went up to 0.9870 at some point

 

My ASIC is 68 and 69. Though this problem with the driver crashing is gone with your vbios, the throttling is also not understandable. We knew before that throttling occured only on high load, now I noticed it around 64%. Wattage is also lower. It never exceeds 70w per gpu. I have't benchmarked the system, but will try later.

Also will do a gfx switch to see if it is better.

So many gimmicks in this laptop. You gotta love them.

Result with no gfx switch
http://www.3dmark.com/3dm/17471123

Temps without fan control peaked at 76 degrees, Throttling occured at 100W power.

Result with gfx switch
http://www.3dmark.com/3dm/17471332

Temps without fan control peaked at 77 degrees. Max power was 104W. No throttling.

With Premas vBios, I have almost the same score but with 122W mas power and temps going near 80 degrees.

I might say that your bios is behaving better. Also no driver crashes. I have revert the power from the control panel to adpative and installed everything from the drivers (physx, HD audio and GFE).

 

If I could get Prema's vbios with 1.0v instead of his publicly available 1.050v vbios I would be totally elated. I love this card, and I love how it's putting out GTX 970 performance but for me I am fighting a losing battle with the temperature, which will get up to and sit between 80 and 85C constant. To me this is way too hot and there isn't really anything I can do, I already have the unit elevated, I have the fans at 4200 RPM, and I have IC Diamond on the core and Shin Etsu heat pads on the memory. Hwinfo64 is showing 140W total power draw. It's my understanding the heatsink was designed with 125W in mind. If I could dial back the voltage and limit the wattage to 120W, even if I lost a few FPS and a few hundred points in Firestrike it would bring great piece of mind if I could get my sustained temperature down around 75-80C.

When I see 85C on my OSD while playing Fallout 4 I begin to get uneasy, especially with some members here, such as TrulyFatal, who have melted the motherboard in their M18xR2 at sustained 90C usage.

http://forum.notebookreview.com/thr...t-take-heat-of-my-980ms.793008/#post-10281763

Any chance we can get this vbios to behave the same as Prema's but at 1.000v / 120-125W? If it involves comparing tables and you need help with it I'm game.

Correction: Prema's vbios is at 1.062v. I would like to try 1.025v and see what that does for the temperature.

 

Have you tried this vbios? I remember on my single 980m and with premas bios, I was hitting 79 tops in fallout 4 with the fans blasting at 4200. Maximum power was 127watt max.

I am using the latest 'mobile' drivers modded by me and utilization for both gpus is around 65 tops. Temps never exceed 72 degrees.

Check again for proper heatsink contact.

 

Oh I'm single GPU, so it's usually 99% utilization in near most every AAA game released in the past year or so: The Witcher 3, Rise of the Tomb Raider, Fallout 4 etc. If utilization drops even just down to 75% my temps don't get out of the mid 70's.

 

I'm trying to sort out the strange throttle behavior guys, really I am. I do not want to use Prema's vBIOS as a base because that would violate unspoken respect rules. I'm not at liberty to release his vBIOS, only he has the authority to do that.

 

I don't understand the continued need for secrecy and propriety at this point in mobile Maxwell's life cycle. It is an obsolete architecture on older platforms with most of the users already running a modified vbios or are obviously happy with factory performance. Personally speaking, if I was Prema, and someone asked me for permission to basically copy my vbios for, lets say, GTX 580M (an exaggerated comparison, but somewhat valid nevertheless) I wouldn't care.

In fact I would just release them out into the wild with my signature on them. So what? This isn't GTX 1080 / 1070 were talking about here, this is an older architecture with a significantly shrinking user-base.

 

Any update here? I will also try a dual PSU when I have the with a wattage meter to see the actual power draw.

 

Raidriar, I am glad to read of your professionalism and your respect the intellectual property of others. This character is sometimes sorely lacking from some in the computer science and engineering disciplines. If we expect others to value our work and intellectual property we much first learn to respect the IP of others.

 

@Raidriar right now I'm using Prema v.2 bios for my twins 980M on M18XR2 but I could use a bios with 1.0v to lower my overall temperatures I tried SVL7 VBios but it isn't working right when using newer Nvidia drivers, so what about your customised vBios is it working on the last Nvidia driver release?

 

I haven't tested, I leave my drivers alone unless there is a fault or a required update. I never solved the random power throttle under max load no matter how I edited the power tables.

PS: How did you get a RGBLED display into your m18x?

 

PM me.

 

Yeah the problem is that this is an outgoing, antiquated platform with a niche user-base, to continue to respect propriety at this point is stupid. We need to stick together and support whoever is left with this platform, to heck with that BS.

 

Thanks Raidriar - my mate has always struggled with running a gtx 980m in his m18x r2 (the continual throttling, shifting between P0 and P5 and having to do the system bios reconfig peg -> onboard -> peg every reboot to work around the issue) but your vbios has none of those problems The only thing he's getting with yours is occasional graphical corruption and sometimes windows boots with a solid screen of colour. I'm guessing this is because of the undervolting? If so any chance you could release one with slightly more volts?

Kind Regards,

Jon

 

Does anyone have the latest modified 980M driver? The latest driver cuts CEMU system memory usage down by 50%; I'm running Zelda: Breath of the Wild with an older driver (378.78) on my desktop PC and RAM usage is at 23 GB, which is ok as I have 32 GB on tap but my M18x R2, which I'd like to prep for a vacation trip, only has 16 GB available....

Edit:

Wow, apparently Nvidia finally got around to officially supporting those of us with 980M:

https://www.techinferno.com/index.p...t-upgrades-driver-support-modded-inf/&page=70

http://nvidia.custhelp.com/app/answers/detail/a_id/4563/~/geforce-hotfix-driver-version-387.98

 

with nvidia inspector up the voltage till you find a stable value and I will make you a vBios with that stable voltage.

For me at 68.3% ASIC quality chip I can run stable at 1.000V. Check you ASIC quality and you will get a general idea of what voltage would be stable for you