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

Valley Extreme HD preset it is!
leave it for how long?
as soon as i can...next few hours.

 

As much as I wanna increase my 880m to 980m swap out chance I do not have the patience to sit through multiple warranty exchanges before I get to my goal. This has become such an inconvenience and iv gone through a doa swap once for a different reason. I'm going to wait for these new cards and hope to get an exchange. If not it's coming out of pocket.

 

Im already throttling and its not even been 10 minutes so I'd say 10 minutes or so. Doesn't hurt to do it longer. I've seen the core dip to 910 already.

Obviously the longer you run it, the better average core speed we will see.

from my HTC One_M8 using Tapatalk

 

This is by far the WORST I've ever seen actually. Its so bad I'm going to remove the dual PSU adapter and try it with single. Wow. Lowest core was almost 780!

Temps were fine as you can see, there's no reason for it to be throttling so much :|

 

I refused to have my laptop back with 880m's. seems it worked

 

my worst wa 663mhz ethrem and that was on a single PSU!

 

Well... same thing with one adapter... except look at the end... the master puked and both cards slowed down but the test kept running... So I'm guessing failure is imminent on these ones too (this is my second pair and I don't drive them hard).



Gonna do Bioshock for a bit more realistic view than a stress test. BRB.

 

Remember the stock vBIOS is slightly tuned to the machine it's being used on. The P570WM could very well have a more proper vBIOS due to the system it's designed for. His 660W input too, definitely means he will never run out of power for the system. Also, the 880Ms have been known to downclock without even hitting full scaling, no? If he's maintaining such a good FPS count and high util count without a downclock, it could very much mean that his vBIOS is sound, even though stock.

Also, BF4 on Ultra might actually keep 120fps constant for him, with his 4960X ripping it up. When I OC my 780Ms to 950/6000 BF4 often caps 125fps on near-ultra or so (depending on how good the server is) without hitting 99% util. And my 4800MQ is a joke compared to a 4960X lol

EDIT: Also, just in case my PSU/Battery have any damage from my voltage fluctuations, I'm ordering a new pair. So I might as well get a converter box while I'm at it lol.

 

I bet he has the same vbios version as me which is why I posted mine. If he doesn't, I want to get a dump of his because this is ridiculous...

I had to reboot to get the cards to even work together again and look at this crap. Bioshock Infinite. Max fans. Max settings. No vsync.

Spoiler

EDIT: Okay, I knew something wasn't right. I used DDU and did a fresh install. Used the 340 series instead of 337. Spoiler tags around the old one to show just how wrong things can go...



Still throttling but its the behavior I'm used to with these cards. No idea what caused a perfectly good driver install to go bad. I'm assuming that when I was running Valley and it puked, either the driver got corrupted or it set some sort of performance restriction on the cards.

Valley incoming.

 

It was said somewhere in another thread... I can't remember where exactly though... that nVidia only gives vBIOS guidelines, and that they make their own vBIOSes. Like I said, maybe Sager has a different vBIOS for the P570WM 880M chips. I also know the cooling in there is better too, so he's also less likely to hit the heat limit of any kind. But as I was saying, your cards (and many others) downclock without hitting a heat/thermal limit, but his appears to not. That's why I called him a special case due to his purchase of dual 330W adapters + the P570WM XD. Hey, for all you know if you get a dump of his vBIOS and throw it in your machine it'd work great =P

 

I highly doubt that Clevo would make a custom vbios for the 880M in that machine considering that they were not even aware there was a power issue to begin with. They used nVidias reference vbios, trusting that nVidia knew what they were doing.

But yeah if sa7ina has a different vbios from mine and doesn't throttle I'd definitely take a dump of it.

On the cooling point, its not that much better than the P377SM-A for the GPUs, using a very similar set up. 2-3C max difference and I've actually seen examples that have them running hotter.

I need to repaste these, its still the stock paste job they put on there when they sent them back to me but I didn't want to even crack it because it's a huge hassle for a few degrees on cards that will just throttle anyway. They run about 2C hotter than my GC Extreme ran.

Here's that Valley run



It fascinates me that Bioshock Infinite actually loads the GPUs harder than Valley.

 

I was running at 1137 on the core as my everyday overclock for gaming. Then I decided to update drivers.

Buh-bye crazy overclocks...

 

Really? What are you stuck at now? And I still can't believe you managed to run that as a daily anyway... insanity. Just goes to show how great the cooling in these machines is when its fixed though.

I don't have any explanation for my driver issue... That was totally random. And makes me nervous... I'm going to have a nuclear meltdown if my cards fail again, I haven't used my laptop in over a week, fire it up today and its being a pain? -_-

 

Here are my 45 minutes results in Valley HD preset:


no throttling.

 

Remember too that the single card machines use Optimus, so the 880M is idle most of the time.

So far 1053 is stable. If I run faster the driver crashes where it didn't used to. It stays below 72c (from memory) when running at 1053.

 

Thanks for doing that.

Nice temps too. Liquid ultra?

Can you check your vbios version for me? I have 80.04.E9.00.10

If your version is different, I would be grateful if you could dump it using GPU-Z and upload it for me.

If it's the same vbios version then I would have to say that the temperature curve is the cause of the throttle and I might need to look into fixing my cooling. That would match my earlier findings when I had CLU on the slave.

That wouldn't explain it all though because GPU-Z shows the throttle reason on my machine being power related.

Its not like the 880M gets a lot of use when it's the only card either though, 0.812v and 135MHz isn't much of anything for 2D.

1053 is still a pretty nice overclock. I'd be happy to have 993MHz without thermal throttling.

Sent from my HTC One_M8 using Tapatalk

 

True enough about the 2D statement.

My card throttles with the stock vbios even when my temps are below 60c.

 

Really curious to get a dump of sa7ina's then.

Sent from my HTC One_M8 using Tapatalk

 

I agree that would be interesting to see. I would like to look at it and compare it to my stock vbios and svl7's missed vbios.

 

That's right, you know how to work with those files.

Sent from my HTC One_M8 using Tapatalk

 

More like lucky guesswork on my part. Ha!

 

Either way haha

Sent from my HTC One_M8 using Tapatalk

 

Hi
as for the thermal compound, Sager put me IC Diamond for the CPU&GPU.

as for the vBIOS mine is 80.04.E9.00.02 as you can see here:


uploaded it for you to my Drive:
https://drive.google.com/file/d/0B-iGj1OnRxA4ZHk1cFRNVmltem8/edit?usp=sharing


tell me if you need anything else.

 

You know I didn't even look at my stock VBIOS version number. My VBIOS is this exact same as the one you pulled from your card. I did a quick check of all of the important numbers that I am aware of and they all matched.

Ethrem, would you be so kind as to dump your VBIOS so I can compare that one as well?


EDIT: Here is an image of the two VBIOS's.

 

Here is my stock one: https://drive.google.com/file/d/0B2uxkA8nKyn6U2Z1Y1NCQjhNRFU/edit?usp=sharing

And here is the updated one Prema sent me some time ago: https://drive.google.com/file/d/0B2uxkA8nKyn6dGdwbHZXT3dPUTg/edit?usp=sharing

 

This one throttles but it definitely throttled less.



It also ran cooler as well.

 

Very cool. Actually, the cards are not running hot. When I used svl7's vbios it doesn't throttle until I hit thermals. I was going to say to try the one from sa7ina and see what happens.

EDIT: You should monitor that run with GPU-Z and see if it tells you why it throttles. If it's "nvidia" or power limits that can be fixed. GPU Shark is a good one too, but you need an extra monitor so you can watch it in real time. It will tell you in plain text why throttling has occurred. You can get more info with GPU Shark if you go to Veiw > Detailed mode


I compared all 4 vbios with kepler bios tweaker and they all appear to be the same. The checksum is different on yours and Prema's so something deeper than kepler bios tweaker can see is going on here which is beyond my limited knowledge.

 

I get the sneaking suspicion that bumping the power up somewhere in the middle between nVidia's setting and svl7 setting would be the key to fixing this excessive heat problem. If those values are right, he's really juicing the hell out of those cards and it may be causing unnecessary added heat.

I'm just not sure about how safe it would be to adjust the power in the tool.

Sent from my HTC One_M8 using Tapatalk

 

Is this from my vBIOS?

 

Yes it is.

The throttle reason was because of the power cap. I'm guessing your cards are different from my own. How I'm not so sure. I'd guess you have better silicon than I do and its using less power. Your cards are still unable to sustain boost with the default power but they do a better job than mine.

If these readings in the BIOS tool are accurate, these cards are hard limited to 117W with the stock vbios. For a card with 125W TDP that's just not enough.

I'm pulling an MSI vbios to take a look and I'm gonna see if I can track down an ASUS.

Sent from my HTC One_M8 using Tapatalk

 

Yes, if it is throttling due to power cap, then it is limited by the vbios. This I know for certain.

 

I think the clock drops you see are from frequent scene changes in Valley.
tomorrow i'll check it in a continuous scene.

 

See? I told you guys his vBIOS would be different.

Also, I said SAGER is the one who did the vBIOSes. Clevo is just the ODM; they make the machine's barebones and design what it is meant to have put into it, but they don't exactly sell the shells with the 880Ms in it XD.

So, see? We uncovered something. It is very possible that the 880M vBIOS in the P570WM is better than the vBIOS in the P37xSM-x models... for what reason we do not know.

Also, as far as I know, Deadsmiley has a single GPU machine. His vBIOS might also be a bit different due to being a single-GPU system, as his seems to throttle far more than yours does, and he also benefitted from the P570WM's vBIOS.

So guys... now we're getting somewhere =D. Why don't we show these examples to svl7 and ask him to delve into the P570WM's as well as the P37xSM-x and P17xSM-x vBIOSes and see what the actual difference is. For all we know, the P570WM vBIOS used as a base might provide a new modded vBIOS that works better for everyone? It may be almost too-little-too-late, but for the people who ARE stuck with 880Ms it might provide a much better working card if they cannot get it replaced/costs too much for shipping to get it replaced/etc

Also I know ASUS' 880Ms seem to run cool as hell though they're 4GB models... they don't sell 8GB 880Ms so we can't compare; for all we know the 880M 4GB work like 780Ms with an OC.

 

Actually, the vbios in my P170SM-A is the same as sa7ina' vibios based on the checksum number. The numbers appear to be the same for Prema and Ethrem's vbios based on what is seen in Kepler Bios Tweaker, but the checksum differences rule this out. I do find all of this very interesting!

 

Double post

 

Hi, I've been watching this vbios stuff and I can confirm with 'm' chips the laptop manufacturer is responsible for setting the variables that control the device. This is done taking into account the machines maximum BTU exhausting capability, the location of heat sensetive components in the confines of a laptop, and finally for legal compliance - they don't want someone burning their legs or worse still a battery spontaneously combusting in an overhead locker!

On my R4 for example, the card is set to go into max P state (75mhz) as soon as the PSU is removed. I can only overclock up to +135mhz on the stock vbios, dell playing safe. The fan tables should also change between laptops. This is why NV have a disclaimer on the drivers. Basically saying if it don't work, don't tell us, see your OEM and use their published driver.

Maybe each OEM will come out with a good vbios for the 880m but history says they just ignore it until the problem goes away, or svl7 fixes it for them!

Good luck guys

 

<S>But as we both saw, checksum being the same still provides two very different performance trains across the vBIOSes =3. Means something can be different with the same checksum =D.</S>

I did not read well enough, feel free to smack me once with minimal strength.

 

Not being argumentative here, but I don't see how I benefited from the P570WM's vBIOS as I never flashed that onto my card. Sorry for the confusion, but I am confused. (Not a difficult task I might add...)

 

Same checksum = exact same file. What you say here makes no sense.

Sent from my HTC One_M8 using Tapatalk

 

Yep that's the whole point of checksum verification.

 

Change the checksum, see if that helps

Spoiler

 

My bad, I was of the opinion that deadsmiley used his vBIOS and saw a change.

 

No worries. I may have not been as clear as I had intended.

 

So I noticed my reason for getting throttled based on GPU-Z changes from either Util or Pwr. It either doesn't boost or underclocks below the 954 stock. Again i know these cards have issues so I was just checking to see if these 2 perf cap reasons are common.

 

Util means that they are being throttled because they aren't being utilized fully. Power means that they are being throttled for hitting the limit on power consumption.

I wonder if these cards are extremely leaky. It would explain the high temperatures and massive power related throttling from what I understand.

Sent from my HTC One_M8 using Tapatalk

 

I just hope I have no additional issues with cooling on my laptop or a bad ac adapter. If its just the card/bios/drivers then ok I will be getting the new card anyways. I actually went into bioshock and minimized the game... watched the core jump back and forth from in the 400's and back to the 900's going from PWR to Util over and over. Card always runs high 80's up to 93 degrees as well.

 

Its always going to drop to util when you switch out to the desktop. Use msi afterburner to have an OSD that shows the core while gaming.

With that said are you on the modded vbios? 93C is really way too hot for stock vbios. The mod will indeed pull the core back to 400 when it hits 94C which is why I either don't use it or I set the core low enough it doesn't pass 90C. For me, it's easier to just not use it. I average 950 core when on the stock vbios and don't pass 82C. It works great for most games but for some, I have to tweak graphics down one notch to avoid having the slow down.

Sent from my HTC One_M8 using Tapatalk

 

I am running stock vbios. I use MSI afterburner to test overclocks and in certain games I can actually maintain an overclock... but those numbers and throttles i'm getting are without any overclock enabled during Bioshock and Metro for example. Most games I play i'm seeing 90's for temps. Only a few like World of Warcraft and Eve online stay in the 70's or 80's.

 

What machine? Sounds like you need a repaste badly.

Anyway turn on MSI's OSD and you can monitor your core and temp real-time.

Sent from my HTC One_M8 using Tapatalk

 

Alienware 17 only got it back in April. Pretty bad if I need a repaste 5 months in. I may just wait to get the 980m card to do anything... Let me jump in a game real quick and see what I get. Btw my idle gpu temp is normally around 60.