980M Blackscreen on 84C+

Have you tried ( johnksss & svl7)'s vbios? It takes care of the throttling/turbo boost, and runs at 1.000v.
No more black screens for me, unless I go above 1.1v & >145W
It's a strange thing indeed regarding CPU+GPU throttling. Maybe when the GPU throttles it bottlenecks the CPU->therefore it throttles down???
I noticed the CPU load fluctuates when my 980M throttles but not the clocks.

 

aw, crappedycrap, checked my pad job one more time, rearranged a little, added a few pads here and there, still no dice. black screen during fs ultra stress test after not even 5 minutes *sigh*

as mentioned before @Prema already provided me with a UV 1.000V vbios, which im on right now. still getting the black screens with that. so i dont think the @svl7 / @Johnksss@iBUYPOWER bios will help

Sent from my Huawei Mate 8 NXT-AL10

 

What about temperatures ? Mine used to do at exact on dot 84C and when it hits 85C (happened only once) , It's imminent like a gunshot.

Really sorry for this, feels real bad when It happens.

 

no issues with gpu core temps, i was at 59C on gpu and low 60s on cpu (4.5 ghz) when it happened

 

Wow, Amazing temperatures !!
90% sure a failing MOSFET or VRM in your case, A professional service should fix it permanently.

 

haha thx, that was with max fans though

and yeah, it points to that direction. im guessing the stock vbios is throttling SO hard that the failing component isnt stressed sufficiently to show its weakness. ah well, it wont bother me in everyday situations, i dont game that often anymore, if at all. nowadays, cpu/ram/storage is more important for me still, it annoys me, especially since i still got 2 more years of warranty on this thing

Sent from my Huawei Mate 8 NXT-AL10

 

mkay, this looks promising: did another pad tweaking, now i finally managed to pass 20 loops of fs ultra with max fans. gpu max. at 58-59C (stock with prema vbios) and cpu max. 60C at 4.5 ghz. score was 97.8%. gpu wattage maxxed around 115-120W (hwinfo64).

now another run, this time with auto fans. wish me luck guys!

edit: well ill be damned, it passed on auto fans as well! hallelujah! now i REALLY hope that this is gonna last just to be sure ill do another stress test for one hour. if it passes that then i can finally let it go... and u can be sure that i wont touch my current pad / paste setup for a long time!

btw, score was higher this time at 98.3 % gpu max at 70C and 127W, CPU max at 75C.

Sent from my Huawei Mate 8 NXT-AL10

 

FINALLY NAILED IT!

one hour FS ultra stress test passed with flying colors on auto fans

gpu max 72C, average 119W, max. 128W
cpu max 80C, average ~72C
score 97.9%

i wont touch this setup anymore!!!!!!!!!

@Prema thanks for your patience buddy, as always much appreciated

 

yeah was just about to say it, our various stock bioses are just fine. They even overclock decently, we actually have OC vbioses when needed, but I've reluctantly mentioned them or shared them, because I believe the stock vbios we use is more than enough. Not everyone saying thier an expert actually is,the way I see it is, if they have a plan to take care of it when the card dies from component death, from OC - like can they replace the components themselves ? If yes, then sure, thats the sort of guy I would encourage to use an overclocked vbios.

 

I have a customer who was speaking of getting blackscreens, even without the driver installed, on one of my cards, but in testing it was fine with me. However - initially I found out I either shipped him the wrong bios a very very old one, or he flashed it himself.

Very odd, first case I've personally had with one of my customers getting a black screen and reporting it to me, I had a guy last year say someone he knew with one of my cards had it and fixed it, but no details were shared - such is the thing that happens - when the issues go away - you never hear about it or from them again, cause people only speak up when they have issues

Anyways , seems the person gone silent after giving them our vbios fix. I just want confirmation that it's the vbios that causes this or not.

 

I believe Phobya for 1.5mm and Grizzly 8 for 1mm are perfect ! usually 0.5 mm almost anything is good at that thickness, but Fujipoly wouldn't be worth it, it's a little too brittle and not as re-usable.

 

I purchased tons of these thermal pads in bulk, all brands, and started offering them for those with GPU needs. Grizzly is my up and coming brand of choice lately, not only thier thermal pads, thier liquid thermal paste and Kryonout are actually great too !

 

The first year of Maxwell II production was very solid hardware. We did 3 months of testing (OCT-DEC 2014) before any public releases (JAN 2015). During late 2015 more and more weaker parts passed QC and then entire last batch of 9xxM has been outfitted with low quality/weak MOSFET's, which could barely run at stock values. They sold the trash last when depleting their stock...

You can take one of those weak cards from the recent/last batches, remove their stock MOSFETs and replace them with the same parts (no need to even add any), just a HQ source and they run perfectly fine afterwards.

 

Do you have / can you share any details on identifying the low quality cards/components from the high?
Production dates, MOSFET id, appearance etc

(both clevo 980ms in my p870dm have dates of Sep '15 and can clock to 1350 at stock premavbios 1.062V so are not the low quality ones)

 

Nope, but they first started popping up during the second production year (late in the second half of 2015, after they attempted to block overclocking on mobile Maxwell via new vBIOS and driver).
When they had more stock of higher ASIC cards (Which use a lot less default voltage), they probably didn't adjust the testing procedure and those weak MOSFET's started slipping through.
[Depending on the specific cards ASIC quality, the default 980M stock vBIOS voltage ranges from apx 1.0120v (80%≥ ASIC) to 1.075v (64.9%≤ ASIC)].

 

Gotta love people resurrecting a thread gone silent for 4 months, just to bash something with baseless accusations and without the proper know how.
Im wondering if it will run at 1.8v?

 

Maybe

Spoiler

BTW, that's a first year production run and still running great!

 

I've returned those Fujipolys, I got fake the package was not satisfying at all, they ship in Plastic packages iirc but the one I got was in a ziploc (cheap one) that too pinned to a cardboard LOL. (Nothing shocking, Amazon it was, had worse than this with 3 Nexus 6's, A GR07 BE... unending)

Phobya are working great as of now & they are original too xD Will try the Grizzly next.

 

Some of those actually come like that. The packaging can differ from reseller to reseller. Modright ones sometimes come like that, and sometimes they come on that same cardboard with plastic case thingy.
The ones sold by Primochill come sandwiched between 2 cardboard pieces. Same for Tyler PC.

Spoiler

 

I think one of my cards has had mosfet failure, just havent gotten around to having somebody replace them. personal problems and all.

 

wow, really ? thats so odd. I ship things in ziplock bags myself, like a TEP module or thermal pads too. I get them in larger quantities and they are already precut though but only with 1 bag from Grizzly. They have smaller bags for regular consumers, maybe this happened for the guy with the fujipoly ?

Phobya come in some cheaper bags, of which are seethrough, too, but always looked a bit sketchy to me, but are the real deal. Why waste tons of money on marketting and packaging when people just want the product!

 

usually small repairs like that can be done easily, and won't cost an arm and a leg. Get it done while the card still worth keeping or selling.

 

This is news to me, almost a warning sign, but where are these cards coming from ? Is this all manufacturers of the cards or just the cheaper companies that do this like DELL or something ?

 

Hey if you don't like don't participate, no need to be offensive man. Sorry for resurrecting the thread, I wanted to reply to it.

 

Would the ASIC% be some method of testing which were the lower or higher quality cards that were dumped with cheaper components ?
I haven't seen anything less than 70% yet from one supplier, and another I have I've seen everything under the sun - but I don;t get my 980M from the latter, only get what I can not get elsewhere.

 

Im not the one being sly and offensive and sugar coating my remarks. Specially with a clear lack of knowledge, if you do not like my comment, feel to ignore.

Specially bringing up a post specifically targeted at something/someones work, again, with a clear lack of know how.

Just like you wanted to reply to it, i ended up doing the same, because i wanted to reply to something you bought up for no reason.



And no , ASIC is not for component quality, it pertains to the cores binning.

 

With the latest batches chances are that lower ASIC cards have higher quality MOSFETs as they run higher stock voltage so their MOSFETs have to pass higher loads during sampling.

 

Thanks for sharing, I will be on the look out and see what I find.

 

Just for reference...I have a woodzstaack card, asic of 78.1 and my mosfet is blown.

 

At stock clocks ? were they cooled properly ?

 

And here I was planning to get a 980M in my GT70.

FML.

 

Is there a way to tell if the card is from an earlier or later batch? Suddenly I don't feel so bad about my poor ASIC rating of only 69%.

 

and suddenly I feel pretty crappy about my 78-79% ASIC hahaha

Sent from my HUAWEI NXT-AL10 using Tapatalk

 

Is this unexpected? 980M is not the top bin product for GM104

Could have been worse, you could have bought a Dell ;p

 

haha i cant complain to be honest i underclock the gpu as much as possible for everyday tasks and only go up to stock clocks for gaming sessions, thats it. my main overclocking focus these days is CPU and RAM

as for Dell: just did some maintenance on my gf´s Dell Vostro 3350 from 2011, still going strong! gotta love socketed hardware while i was at it, i ordered her a 7260 ac to swap out her crappy 1030 N wifi adapter

 

I see your 2011 dell and raise you a couple of Core 2 models from even earlier... The Vostro 1500 in particular is immortal, what its been through over nearly a decade ought to have killed a brick. But waaaaah waaah a 4kg magnesium frame in a 15 inch is too heavy, say your average idiot consumers......

/OT
In case it wasnt obvious I was referring specifically to the history of low quality dellienware top end GPUs... In contrast to their price premium

 

Well, that does sound rather heavy given that the 17" Precision M6400 of the time weighed less (like 3.8 kg or so). Still works as well, BTW.

 

haha i hear ya boys. for a 13 incher her 3350 is by all means quite thick by today´s "standards". it originally came with a 250GB HDD/Core i5/4GB DDR3-1333 and i upgraded it to a core i7 2760QM/850 Evo 500 GB SSD / 8GB DDR3-2133 HyperX. Now the only things left were the wifi adapter and (potentially) the screen, but not sure if i can find something higher res that would fit the chassis.
unbeknownst to my gf, i now applied CLU on her CPU and shes amazed at how cool and silent it now runs

 

I found out that my recent buy of Dell Alienware 980m has this Black screen issue when i push the clocks too far. 1220 Mhz boost clocks for the core are working fine, but if i push above that i start getting the black screens regardless of temperatures. 74 Degrees temps with max fans.

Installed new Phobya Ultra 2mm 5mk thermal pads on the mosfets, didnt improve at all from stock clevo ones.

Still happy with 1220mhz boost, but temperature-wise still has room for more clocks.. i hope it doesnt start degrading...

My asic is 76,1%

 

The reason of Black Screens in 980M is the quality and quantity of mosfet/vrms. Since GPU's doesn't have installed sensors, we can't monitor their temperatures.
The only way to get rid of black screens is to hard mod the GPU... which will void warranty and might be dangerous to card itself if done wrong. But if made well - it will improve GPU's lifespan (lesser loads on mosfet/vrms, better voltage stability etc).

Thermal pads won't help much. Already tested Grizzly Minus Pads - it has been improved a bit (20-30Mhz higher clock on same GPU) but... overall waste of time.

 

Hello all!

I have the same issue with my 980m card on my AW17R1 Ranger, I'm using Prema v2 BIOS on an eBay card I bought and the voltage is set 1.067V and I get high temps (84-85C) while gaming. When the temp reaches 85 I get the black screen and have to cold boot my machine. If using inspector I throttle the temp to 83C the machine is rock solid. Is there a BIOS mod that can get the voltage to 1V from which I hear it has very good results on overheating. I do not want to overclock or anything, just to be able to game without issues! I have also repasted with liquid metal Thermal Grizzly Conductonaut which dropped the temps from 90-92 to 83-85 on load and from 55 idle (with fan on) to 47 with the fan not spinning.

 

Check VRM thermal pad contact.


For undervolt, yes you normally need to undervolt via modded flashed vBIOS (or find one someone's already done). You can try the voltage slider unlock procedure for Afterburner, but that usually is only able to overvolt. 1.06V was enough for the 980Ms in P870DM to run a 1375mhz core OC and IIRC they were just under 80% ASIC so yes you could undervolt and it would reduce power & temps (and stress/heat on VRMs) nicely.

Yes, usually the VRMs lack a heatpipe over them so if the heatplate is heating up too much even the best thermal pad will do little to lower their temps, provided contact is good and the pad isn't old and dry. That AW17R1 heatsink looks relatively poor for VRM cooling, very little metal mass (aluminium too?) over a fair distance connecting VRM area to heatpipe. (For comparison; the P870DM heatsinks that could sustain 140W at 75C on liquid metal on 980Ms were all copper, and had a 3rd heatpipe over the memory portion of the heatplate. No part of P870 was engineered with weight reduction in mind haha)

 

The problem is I cannot undervolt using inspector... I'm in need of a vbios without OC. I have tried some but most of them have stability issues... If anybody can help I can send you any details needed!

Thank you!!!

 

my cards settings

 

What is your AISC value?

 

It is 66,9%

 

You would need a vBios with at least 1.006V at stock clocks for this AISC. So a modified vBios like SVL7 wouldn't work for your because it runs overclocked at 1.000V.

I can make you a stable undervolted vBios, if I remember correctly I have seen you at AW sub-forum. You have an AW17-R1 right?

Naturally you shall use it at your own risk. I am not responsible for any damage that you may cause to your system while flashing the vBios.

 

Yes I use an AW17 R1 (Mid2013) with A14+ modded BIOS. I really appreciate your help and be really obliged if you did that for me!

 

Here you go. This is based on stock Dell 980m vBios. People here usually prefer Clevo vBios, but honestly from my experience (Tried Dell, Clevo and MSI) the Dell one works the best on the 17R1.

If you already know how to flash then happy flashing! if not you can always ask

 

@MahmoudDewy Thank you! I'll gine a try this afternoon and send you my feedback ASAP!