Pidge from Nvidia has asked that user experiencing problems with the 880m list them here..

it is on high,so my scores are under normal 780ti,not overclocked?fail xd

 

I'm baffled by your scores. Your clocks seemed so much more stable than mine yet my stock benchmarks are across the board higher.

I'll have to run some tests when I get my machine back... whenever that is... -_-

 

i hope u wont have to would be nice to finally see this fixed.

 

Wow, there's so much noise in this thread!

 

Good point! =D

And yeah there's lots of noise in here, 880M owners like us may finally get a fix!

Sent from my HTC One_M8 using Tapatalk

 

Sorry if I am silent at times. I get pretty swamped with other work so sometimes I might not be able to provide regular updates unless there is some change in the status of the issue.

 

Of course, your priority is the GTX 980M. :thumbsup:

Those drivers that deal with increased L2 cache and its agressive usage to cope with the 256bit bandwidth to make it shine on higher res than 1080p and have a fighting chance against the 384bit 780 Ti have to be made. They are not created out of thin air you know.

If you let me have some samples I can help you with that work. Maybe do some promotional for Nvidia while we are at it. I have a working AW18 too you know, so you and me can test SLI too. Maybe do some teasers for the folks, sort of a preview on what to come.

:yes: Y/N? :yes:

The NDA protocol being used across the reviewer sites is so boring. Everything is all silent, nobody knows anything, then suddenly *BAM*, 1 million sites writes about the same thing. yay not.

 

HAHHA!! Your probably the last person they'll give a 980M to to test ...

 

Oh well, he should know by now that we plan to post whatever benchmark, GPU-z etc tidbit of information we can find anyway. They should control the leaks and give instructions on what would be allright to post instead.

A weekly update would be cool. Picture of the VRAM one week, picture of the die the other week, power supply the third week, some parts of the GPU-z, etc etc. Could make that pretty interesting imo.

But I know thats not happening anyway, so we will have to go through the asian route as usual. The guys who get the chips long before Nvidia announce the cards.
Like the 600M series.
http://forum.notebookreview.com/gaming-software-graphics-cards/649052-hurray-nvidia-600-series-not-just-fermi-kepler.html

 

by the looks of it they have a fix and now we ar waiting on the oems to drop it or adopt it..

 

What makes you say this DumbDumb? You cant give us hope like that and not give sources/references/links That's just cruel!

 

https://forums.geforce.com/default/...-for-a-nvidia-response-/post/4278730/#4278730

That's Pidge's statement on the issue. Now we wait

 

i think there wont be 100% fix on this without sacraficing something else sadly...

 

I would take a few degrees more heat to get my cards performing how they should.

Sent from my HTC One_M8 using Tapatalk

 

Interesting - I wonder if this will fix the super weird utilization spikes I get in some titles too (BF4 and Watch Dogs).

 

No, those are just poorly optimized games. The game developers have to fix that.

Sent from my HTC One_M8 using Tapatalk

 

Sager opted to replace both cards, not just the defective slave. I'm intrigued. Machine will be here Thursday and I'll report back.

Sent from my HTC One_M8 using Tapatalk

 

You misunderstand - my BF4 will have stuttering due to the GPU dropping to 40% utilization for a second or two for no reason. It happens constantly. I can't get a locked 60FPS on any settings, regardless. None of the drivers have fixed this. My FPS is getting halved and then climbing back up before starting the cycle again.

 

I do understand. Those games have terrible optimization. It's not a GPU issue. My desktop 780 Ti does it with Watch Dogs.

I don't play BF4 so I can't comment on that one but there are reports all over of performance issues with it

 

NICE FIND... but not what i wasreffering to.. I mean no offence here meant but.. what are the odds of a nvidia guy having a flawlessly operating 880m.. come on read between the lines.. this seems like nvidia knew this was an issue and hoped they could skate past until the new cards drop and not have to do a thing..
He is the only person i know that has a PROPERLY WORKING 880m.. some of may say ares work fine but not like his.. He had absolutely no drops/spikes down time at all..

im just glad he is pushing the issue for us.. he is under a NDA and cant tell us if his vbios is new or what.. theres alot he cant tell us.. but im just glad he is helping..

 

Ummm... Pidge had massive throttling, I think you got confused somewhere. His brand new Alienware 17 was hitting 94C and the core was throttling below 770 because of it. The dude that doesn't have throttling doesn't work for nVidia and looking at his benchmark numbers, they aren't as high as they should be.

Sager put two new cards in my machine, hopefully they are July cards so I can see if it was early silicon issues that were a big part of the problem with the high temps. I'll know tomorrow. Seeing 968mv under 3D load makes me hopeful, especially since both of his cards were like that.

 

in case u are refferning at me,my cards dosent throtle ingames,played whole crysis 2 with no throtle at all on 993 mhz but in benchmarks oh yeah baby they do for some reason unknown to me.idk maybe all of u throtle just in benchmarks but what i read so far it dosent look like that...crysis 2 should be some kind of performance test for gpu or not?

 

Yes I was referring to you.

I don't know about Crysis 2, never played it. Bioshock Infinite really seems to be the best test for burning up the GPU from my experience, its very well coded so utilization is always between 95 and 99% and without vsync it will heat the cards up very fast.

Sent from my HTC One_M8 on Tapatalk

 

OK, then there's potentially something wrong with my card as the GPU utilization graph had huge drops for a second before correcting itself. I assumed this was to do with streaming issues in UE3 but maybe not? I do have v-sync on so it's not going to stay at that high a rate which might explain something. Then again, the DLC had a propensity to stay at like 57/56 FPS for prolonged periods (with 70% usage) for some reason.

 

Vsync always causes weird issues which is why I use the adaptive setting in the nVidia control panel as opposed to native vsync in the game.

I'll have to make a utilization graph when my machine comes in today.

 

Hmm, I'll try that. Forgot that they finally let mobile users turn on adaptive v-sync again. That was removed ages ago and I never bothered to check whether they'd brought it back.

Maybe that'll explain my WD/BF4 issues too? Shame I uninstalled the latter soon after testing it - the problem with only have a 256GB SSD in my secondary....

 

I don't think Adaptive V-Sync was ever restricted on notebooks per se, you just can't use it if you have Optimus. He has SLI which means no Optimus.

Adaptive VSYNC missing from notebooks with NVIDIA GPUs featuring Optimus technology

 

Yeah this is correct.

 

That's pretty common with BF4... not sure we can blame that on 880M entirely. It doesn't seem to be related to GPU performance. It happens on systems with both weak and powerful GPUs to one extent or another.

 

Well damn. That's annoying. Even if I have it set to default to the Nvidia card? Boo.

 

Got my laptop back today... Two new cards, still no good.

Unigine Valley

Master:

992.9 - 1.012v
979,8 - 1.006v
966.8 - .993v
953.7 - .981v

Desktop 135 - .806v

Max temp: 76C

Slave:

992.9 - 1.018v
979.8 - 1.006v
966.8 - .993v
953.7 - .987v
940.6 - .981v
928.1 - .968v

Desktop 135 - .812v

Max temp: 79C

vbios: 80.04.E9.00.10

Same crap as before, pretty much exact.

 

both of mine are at 0.800 on desktop but it happens to me also that on desktop mode for some short period of time i see my slave being 2-3 degrees hotter then master but it goes down in par secs.. my vbios is 80.04. E8.00.10

btw are u using ur laptop on table or cooling pad or something else to make more space for air to get under ur laptop?


also u could have wrote down ur asic quality and compare it with ur ''new ones'' just to be sure u got new stuff xd

 

Your vbios is E8, not E9? Hmmm.

I have the back elevated using two pill bottle caps and an ambient 21C temp.

It's weird that your cards are both using the same voltage, just doesn't sound right. I monitored the voltage and clock speed in real time using RTSS and Hwinfo to have an OSD and wrote down the values. When you are in GPU-Z on the desktop and click the drop down box to switch between cards, they both settle down to 0.800v?

Sent from my HTC One_M8 on Tapatalk

 

ya man,i open gpu z twice to monitor both cards in real time,both are locked down at 0,800v this ofc change when gaming,master one is slightly under bigger voltage as slave but u know that already.seams bios u got is new one i guess..i dont have it and my laptop is like 4 weeks old.or alienware uses this one still that i have,wouldnt know..go check do they throttle xd


to make this clear , i have my laptop on table,but i can pull glass out and my laptop is brighter as hole so it has planty of airflow it helps to keep it cool,at least 5 degrees if not more when gaming.

 

My Steam Library is copying over from my desktop right now, once that is done I'll see if it acts the same in Bioshock Infinite but I have a feeling it will have the exact same results as before. I have no way of proving that they didn't replace the cards other than the voltage being the same unless the slave card pukes again.

I think I'll just deal with it for now, hope Pidge posts a fix, or start putting money away towards a single 980M or a pair of 980M depending on the cost and performance and hope that Prema releases a BIOS to support them.

 

Here is the problem with editing posts... I missed this.

They are new cards. It appears that the vbios policy is what is causing the issue.

Master is 81.6, slave is 80.8.

Old cards, master was like 82.5 or something and the slave was 85.something

Thanks for that, I thought it was kind of a stretch that they would swap the stickers and flash the vbios rather than just saying there was nothing wrong with the cards and shipping the machine back but there are companies out there which would do this, glad Sager isn't one of them.

I have no idea why yours are using lower volts than mine, once this is done I'm going to try to undervolt them (stock vbios, I'm not messing with that modded vbios) and see if the performance drops or 3D crashes, etc.

EDIT: Undervolting does nothing as I suspected, thanks boost! -_-

 

would be nice if others post their voltages also so that we can see whats average V of this cards

 

This is interesting... applying a core overclock to the "weaker" card dropped the voltage... The card is definitely a step weaker than the master as when running the same voltage the master was, there was some flickering on the screen on Valley but that's an interesting finding either way. It didn't actually go to the higher boost clock, it just equalized the voltage at a given clock between the two cards.



The bottom is the weaker card.

Unfortunately the throttling still happens. I'm baffled.

Edit:This is so weird... I have been experimenting and +15 on what is supposedly the weaker card drops the voltage and even pushes that card to 1006MHz occasionally and it isn't throttled below 953.7 often and when it is, it goes back up quickly and that's with Valley which is supposed to throttle. I'm gonna fire up Bioshock and see what happens, I'll report back with graphs.

Any setting other than +15 and it still drops to 928...what on earth did nVidia screw up?!

Sent from my HTC One_M8 on Tapatalk

 

This is what it did until it froze up and crashed. Zero throttling below stock 954MHz.



Unfortunately, it went to a black screen after this and I had to turn the computer off and back on.

It does give me hope that this can indeed be fixed though and as you can see, the voltage wasn't crazy. It also went to 1006MHz for a while.

Here's a stock Catzilla, no CPU or GPU overclock: 13495

These cards are indeed performing better but they still need some work. At least now its bearable waiting for the vbios update.

Here's a stock Bioshock with borderless window to stop the freezing



Things look much better there. I can be happy with this for the time being. Off to go play now!

 

Yes, if you have an Optimus system, you're SOL as far as Adaptive V-Sync is concerned.

 

Sorry this is probably a bad question. But my np9570 is coming within a week and i definitely want to get my 880ms running great as soon as i can!!! So, i have a question for you good sir Did you simply increase the voltage by +15mv? and then the card runs stable at the correct clock speed? Why did it crash for you? Overheating?

Thank you very much!

 

Unfortunately it crashed and now it won't do the +15 on that card but it doesn't matter, I threw maxed out sleeping dogs at it and was rewarded with 80C on the master, 82C on the slave, 954MHz master with the slave dipping down to 888MHz at one point.

There just isn't any way to get these cards working properly with the stock vbios and the modded vbios will boost til they pass the throttle zone but in my experience, the memory gets throttled at that point, not the core, which drops the performance quite sharply (memory gets knocked to 2000 immediately, a nice 500MHz drop). There's also the issue with the cards getting stuck on 3D clocks causing the fans to run all the time and the cards to idle mid 40s lower 50s.

Basically, these cards are broken. Even with a modded vbios, they still don't work right. The only hope we have is nVidia fixing the stock vbios and releasing it to us.

With that said, these cards are stupid fast and I can only bring them to their knees with insane graphical settings that you would only notice with a magnifying glass. There are exceptions though, like Watch Dogs, but that's just poor optimization.

Sent from my HTC One_M8 on Tapatalk

 

I also confirmed with single GPU that the master will throttle to the same core that the slave was doing with SLI so it's definitely a vbios ceiling.

Sent from my HTC One_M8 on Tapatalk

 

You rock Ethrem! Do you think Nvidia will continue to "work" on this issue even after the 9xxm series releases?

 

I have faith that Pidge will get a fix out, the question is how will it be distributed, vbios flashes are risky business. I hope they don't just send the vbios to the OEMs and leave it up to them to deploy it.

I really don't want to buy new cards and besides that, Clevo bios has to have support for the cards or the machine won't POST. Prema adds new cards to his modded bios so that isn't the end of the world but its just another thing to worry about plus the new cards wouldn't have the warranty that my machine does.

Sent from my HTC One_M8 on Tapatalk

 

I see Asus using nvidia gtx880m with only 4gb of VRAM on the graphics card. Does anyone know how they is compared to these problematic gtx880m with 8GB VRAM? :hi2:

 

They're most likely just slightly hotter than 780M cards although I'm not familiar with ASUS and their cooling system in comparison to my own.

Sent from my HTC One_M8 on Tapatalk

 

Hello NBR Community,

first of all i apologize if my english is not perfect, because i´m from Germany

Since May i own an Alienware 18 with 2 GTX 880m in SLI. After several attempts to optimize my temperatures (jump easily to 90 degree with full load) i don´t want to deprive / withhold (dont know which is the right word ) my results.

In Battelfield 4 with full Details and 150% resolution scale my cards easily hit 85 degree+ when i got my notebook. With my optimizations at stock core speed the temperatures hover at 71 - 72 degree now and the cards stay at 993mhz all the time.



here are my steps to reach this temperatures:

- first of all flash SVLs MOD VBios. Pls expect that your stock cooling system with stock thermal paste can`t handle the massive power of these unlocked cards. The card will hit 93 degree very soon and throttle.
- lift the rear of the notebook. I use 2 little rubberfeets. The fans can pull the air better now.
- repaste the card. I use Liquid Metal, but be carefull with this.
- The MOD Vbios allows you to undervolt the card even at stock speeds. Be aware that every card is different. My cards run stable with 0.981v at 993mhz. Your cards could run better or worse

on very warm days i hit 90 degrees, even with my optimizations.


So this is not the final resault look at this:



after setting the core speed to 900mhz and overclock the Vram to 3100mhz my temps getting lower again, but my performance increase 10% . On warm days the difference could easily hit 10 to 15 degree (28 degree room temp -> anno2070 -> 92degree vs 78 degree).
With 900mhz core speed im able to run my cards at 0.918v


I hope some of you will test this and i hope it will help alot to get rid of these high temperatures

greetings from Germany

captn1ko.

 

Thanks for sharing. It's irritating that we have to take such extreme measures to keep these things somewhat cool.

I played with undervolting mine and unfortunately they can't handle under 12.5mv undervolt on both cards (how did you get 0.981? My program only gives me 12.5 increments).

I found that with stock volts the cards will hit 90C with 98-99% simulated load with 3dmark but in games, they run 993MHz to 88C with auto fans. It's hot but decent.

I don't like liquid ultra and won't mess with it again. It's easy to put on but if you take it off, you have to be really careful as it likes to fall off the qtip in little beads... No thanks.

Sent from my HTC One_M8 on Tapatalk

 

thats right. You have to be realy! careful with liquid Ultra... But it does a great job in combination with the other steps. The point is: with ur 88 degrees your fans kick in realy hard. My fans run on second level after idle (then my temps hover around 77 degrees). Thats a huge difference (for me)