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

I'm more inclined to think it IS some sort of vBIOS guideline or some kind of error as such... the 870M cards have 6GB vRAM and don't get *NEARLY* as hot, so 2GB extra shouldn't destroy a card's heating regulations. I definitely feel it's something on the card or on the components itself, rather than the actual intended design.

 

it could all be software related.. or it could be flawed components.. my machine shuts off before any thing bad happens.. and it is not from temps.. because i could start a game and blam it shuts off.. and it might run just fine the next time..

Dell or Nvidia for me needs to respond toi this thread or some one with a good rapport with them needs to try to get the skinny on this.. we need just one company to give us the whats what about these cards..

but in games it does throttle but the temps never get above 70-78c on mine.. and there oced. they just dont ramp up..

 

If your PC black-screen shuts off randomly, it could be a memory issue. I suppose you would have likely checked this already, but I can't rule out the 880M's vRAM doing it. But it sounds like something my D900F did at the end of its life, and it was random. Temps were not the issue; definitely.

 

yea not a memory issue as ive tried a few different sets..

 

It makes no sense to say you have it overclocked but it throttles. If it's throttling you're on the stock vbios and that can't be overclocked successfully. It will actually cause the cards to crash (not enough voltage for much over 993MHz, 1006MHz was the highest boost I got with the stock vbios and it throttled faster with the overclock) and bumping the voltage up will make them hit the power target and you won't even hit stock clocks.

If you are running the modded vbios and your cards are throttling, you did something wrong.

Sent from my HTC One_M8 using Tapatalk

 

really how so? because ive learned from the best.. stock.. oced..modded or unmodded bios i9 throttle. i also get the screen tearing line across the screen that stays about right in the middle.

 

Then you have bad cards.

My cards do not throttle with the modded vbios at all until they hit the thermal throttle point. I can show you some graphs if you'd like.

If the temps didn't go so high, the cards would be perfect. I set them at 849MHz @ 0.893v and they don't budge from their clocks and they stay under 80 now.

When the mod does throttle, its bad. I'll drop from 993MHz to 405MHz which makes everything stutter for those few seconds then it goes back to 993MHz until 94C again. It really blows.

Sent from my HTC One_M8 using Tapatalk

 

lol every one has bad cards.. some work better than others but in the end the 880m was a huge flop and role of the dice when it come to performance.. some times mine work just fine and other times they do not.. just because your problems are not that of every one elses doesn't mean there not there or one simply has bad cards.. it has been proven all ready that peoples hardware will act differently but ultimately come to the same results.. now how can that be.. you don't throttle until a certain temp is reached.. me i throttle at random.. others have different issues that all lead to them throttling and other issues.. arguing back and fourth and throwing blind solutions out there is not going to help any one.. we need to band together and get on nvidia and are respective manufacturers to get answers.. on the new 18 it doesn't help that every thing else is throttled..

 

Fact: if your cards are throttling on the modded vbios and they aren't throttling because of heat, you should get them replaced.

svl7 made the modded vbios to stop the stock throttling behavior by disabling boost. If you are core throttling still, either you had a bad flash or you have very defective cards. This is a fact, not up for debate.

The rest of us are ticked because the cards run like molten lava with the modded vbios that stops the core from throttling until 94C which means they hit 94C eventually and then go into extreme throttle mode to protect the hardware.

We are all complaining because the cards run so much hotter than the 780M at the same clocks and because the stock vbios is garbage but none of us but you have cards that still throttle with the modded vbios unless it's because of thermals.

Get the cards replaced.

Sent from my HTC One_M8 using Tapatalk

 

Actually, my 880M throttled with the modded vbios until I tweaked it for more power.

 

How so? I didn't have any throttling with either set of 880M cards I've had with the mod until it was thermal. Did you have to increase the voltage or what? Default power target was plenty for me, didn't touch it.

Sent from my HTC One_M8 using Tapatalk

 

My card was throttling due to reaching the power limit even when I raised it to max. So I tweaked the modded vbios to raise the limit beyond that provided.

 

What was the core throttling to? My cards hit the power limit but it's enough to keep the core from actually throttling in response so I never messed with it.

Sent from my HTC One_M8 using Tapatalk

 

I will have check my notes later to see if I recorded that.

Watching the season fanile of Falling Skies with the wife.

 

Word, I'm just curious because I've gotten my cards to 1006 @ 1v with no throttling despite hitting the power target and that was stressing them with the first scene of Fire strike maxed all the way out on a loop. Hit 94C in 18 minutes but didn't throttle before that lol.

Enjoy the show.

Sent from my HTC One_M8 using Tapatalk

 

Well, my 3 pages of notes weren't great but...

After my cooling mods, including sanding my heatsink flat I had the modded vbios installed. My max temp in 3DMark 11 (score: 8875, driver 327.23, no Geforce Experience) was 74c with the core at 993, memory at 1475. Since my temperatures were under control I decided to push it.

3DMark 11: 9525 core 1114, memory 1475, 1.050v 86c
3DMark 11: 9605 core 1143, memory 1475, 1.075v 90c, power limit reached, mild throttling (throttle not recorded)
3DMark 11: 9597 core 1143, memory 1475, 1.075v 88c, Added +30% to power level, power limit reached, throttled to 928MHz at the end of the first scene
3DMark 11: 9581 core 1143, memory 1475, 1.075v 90c, Added +60% to power level, power limit reached, throttled to 928MHz at the end of the first scene
3DMark 11: 9637 core 1143, memory 1475, 1.075v 88c, Upped power level in VBIOS by roughly 20%, power limit NOT reached, No throttling
3DMark 11: 10013 core 1170, memory 1460, 1.100v 89c, Upped power level in VBIOS by roughly 20%, power limit NOT reached, No throttling

I wish I had kept better records, but I didn't. I did save my tweaked VBIOS. I am currently running version 6 of the tweak as I am still playing with it.

 

Well yeah you'll get a throttle if you overclock without raising the power limit but the way Dumbdumb made it sound, he was getting throttling with regular boost clocks on the modded vbios. He didn't specify throttling with an overclock, I assumed it was stock.

Interesting that you had significant throttling with those overclocks though. I don't believe I ever ran into a throttle zone, I'll have to test it one day now that I have dual PSUs

Was this like a hex edit of the mod you did to fix the power limit?

Sent from my HTC One_M8 using Tapatalk

 

How did <del>your casing</del> the MXM slot not totally meltdown from all that <del>power going to its head</del> awesomeness pumping through its copper veins

Assuming the modded vBIOS started with 125W...

125W x 1.3 x 1.6 x 1.2 = 312W

WUT

(yeah I know my math must've went wrong somewhere...)

 

No, nothing like that. Svl and Johnksss did all the heavy lifting for us. I just used Keplar Bios Tweaker on the modded vbios and played with the settings some. It was a lot of trial and error, but like I said... others did the REAL work for us, and we owe them a big thanks.

 

Well you're right, at such high overclocks it throttles. My cards unfortunately don't like 1114 even @ 1.087v so I won't be able to test much further but right before it froze the slave dropped from 1114 to 1108.

Sent from my HTC One_M8 using Tapatalk

 

Honestly, I have been reluctant to spill the beans on this. I don't want it to seem like I did something extra ordinary, when I haven't. Others did all the hard work. I very well could be screwing up my system but I haven't so far. Cooling is the key. Max fans, sanding the heatsink flat (AW has thicker plates and are pretty good already), and monitoring temps.

The GPU heatsink in the single GPU Clevos are split between the actual GPU die and the rest of the components. This is probably why they cool better than the SLI Clevos? I say this in regards to the 880M with the uber hot components (R22 I think it is, from Johns post you linked).

Spoiler

I have been playing MechWarrior Online for the past 45 minutes or so. Temps reached 79c with the core locked at 1150Mhz and 1.100v. I do cap my fps at 70. Not sure how hard this game is on the GPU though. It's pretty much all I have played that taxes the machine at all.

Spoiler

 

Such numbers... 3dmark 11 graphics - 19567

Http://www.3dmark.com/3dm11/8671856

It took 1v on each card and the slave hit 90C (master was 86C) but other than a blip in the first test where it slightly dropped, it maintained 1114 throughout.

Fire strike graphicss: 13309, max temp 79C on the slave

Http://www.3dmark.com/3dm/3935131

No throttle the whole time.

I am going back to my comfort zone now lol

Sent from my HTC One_M8 using Tapatalk

 

So, what did you change?

 

Core @ 1114, 1.1v

Running fire strike again with +300 memory just to see what an impact that makes.

Sent from my HTC One_M8 using Tapatalk

 

Woah, the PCH was hitting 98 and averaging 95 there. Those things have an upper limit of 100c-105c, you should probably keep an eye on it or maybe try a cooling mod for it? I am not sure how the P177SM-A has its PCH cooled; I know the P37xSM-x have it just open on the motherboard apparently. I think my PCH hits 90-92 sometimes, but once I managed to make it cap out at 102 degrees by... using my system to maximum potential XD.

For future reference, do NOT (at the same time) x264 render a video in handbrake while watching a livestream on twitch.tv while playing Black Ops 2 borderless windowed at 1080p max graphics at 120fps constant on a P370SM3 if you want that PCH temp to be low

 

Yep, it's always ran hot. It has been bothering me a little bit. While high temps on the PCH cause file corruption? I think I am seeing some of that.

 

The PCH is essentially the South Bridge, but I do not know if it can cause file corruption. I've never had that be an issue before. Usually mine doesn't pass the temp of my CPU though; only when I am using my CPU and GPUs to their utter limits. Livestreaming doesn't even do it, I need to be doing more. It's why it surprised me yours got so hot, and why I mentioned looking into possibly a small cooling mod for it. I think one was to (assuming yours is under the keyboard like mine) put a small thermal pad on it and put the keyboard on top of it; but we don't know if that has any long-term implications on the keyboard itself.

 

I think there is a mod for this system. It involves cutting out some plastic so a thermal pad can be placed on top of the PCH so it contacts the bottom of the keyboard. I doubt that the keyboard would be harmed. The keyboard is the graphics heatsink on my son's T420S. Not an issue.

 

Crap! I just took another peek at HWinfo64 and my PCH has reached 108.5c. I need to look into that...

Here is a link to n=1's post on the P37xx series mod.

http://forum.notebookreview.com/sager-clevo/755540-5-mniute-mod-improving-pch-cooling-p370sm.html

 

The GPU heatsink design is pretty uniform across all Clevos -- 2 pipes to core + 1 pipe for vRAM and edge VRM circuitry. Maybe you won the 880M lotto or something. But seriously, 312W of power running through the MXM slot and 880M power circuitry, that just leaves me speechless, and at the same time amazed that nothing blew or melted.

Tjmax for the PCH is 108C according to Intel, but Tcatasrophic is 110-128C. It's on page 276 of this Intel datasheet.

So yeah definitely keep it below 108C. I'm assuming Tcatastrophic is given as a range because above 110C damage could start to set in, although Intel does state the "catastrophic trip point is 120C", whatever that means. Either way I'll take 110C as the "never exceed" temp just to be safe.

 

Good to know! I could not find the Tjmax at all for haswells.

 

I have the laptop torn down at the moment. The PCH is covered by plastic. Tomorrow I will remove enough to stack up some thermal pads so they will contract the bottom of the keyboard. Getting sleep and can't run the dremel in here at 2am.

 

Yeah those temps are absurd.. I barely hit 85C on my PCH when gaming and hell I have a 3940XM running at 3.8GHz during gaming so these temps are shocking...

 

I got it squared away. Cut a hole and added thermal pads so they sandwich between the PCH and the bottom of the keyboard. Haven't seen anything above 78.5c on the PCH today while gaming. The keyboard is definitely warmer. I think it's a good trade off. Maybe I can come up with some other way to cool the PCH later.

Ok, enough thread derailment. Back to the topic. Nothing from nVidia on the 880M issues lately?

 

if we hear anything from them, I imagine it won't be until after 980M has been hard launched. NVidia is probably putting all of their resources for mobile into 980M and 970M right now to make sure they don't end up with another problem child.

Sent from my HTC One_M8 using Tapatalk

 

pics or it didn't happen

 

I will make another thread for that. I took pics just for you. LOL

 

LOL

I'm sure it'll benefit all the P170SM-A owners who wish to do something about the PCH temp.

 

I do not need to modify anything on my laptop, when my pch temp is 50 max 65 degrees celsius

 

LOL... Your lucky

 

Ok, done!

http://forum.notebookreview.com/sag...-5-minute-mod-improve-pch-cooling-p17xsm.html

 

Since this is outside of the Alienware sub-forum I have no moderator permissions, so I will have to quote instead of copy... very relevant for this thread.

 

I really hope that nvidia knocks one out of the park with the 900M series.

 

That would be nice, but it would be even nicer for them to take ownership of their mistakes and make things right for 880M owners. The " better luck next time" approach is unacceptable.

 

Beware your hopes or Nvidia decides to skip your luck next time either.

 

So, after being away for 2 weeks, I guess there's no fix incoming? Have browsed the thread and can't seem to find anything.

I'm still mostly happy with my performance but it's a shame to see that things aren't as they should be.

 

Nope... our ball sacks are left flapping in the breeze after nvidia kicked us in the bare nads.

 

On a side note, XoticPC has some leftover NP9380s with a single 780M and 120Hz screen. That is what I was going to buy the day they were pulled. Bugger...

 

Hahaha.. thats gold! :laugh:

Thanks for the laugh!

 

Well, guess I'll be selling this on ebay or something as soon as the 980m releases. Shame.