thread created on nvida forums about the 880s and the bs that there doing.. go reply!

I'm not going to spoon feed you Cloudfire. If you haven't found them you simply are not paying attention. You did not even pay attention to my reply or you would have realized where you went wrong here. The trouble is that you found the ONLY example I am aware of--using my numbers--of a system with 880M that works well. I chose that system on purpose because, YES, that example seems to work right (when it wants to... it still has problems if you are paying attention to everything the owner has to say about it). You are apparently attempting to look like an authority on something without a proper understanding of the discussion, or the problems many are facing with 880M cards that don't work correctly. If you are going to tell everyone that owns a system with 880M that is not working correctly that they are stupid and don't know what they are talking about, feel free to be my guest. As you have astutely discovered, Mr. Fox's benchmarks prove that 880M runs great, if and when a working example is identified. Congratulations, and thank you Captain Obvious. Too bad not everyone has good samples.

 

Even my "good sample" isn't stable with the modded vbios that fixes its performance issues, freezing often, and the stock throttling is insane.

Calling these cards broken is perfectly acceptable.

... so is calling my 4940MX broken...

Sager's 15% restocking fee is looking like it might be acceptable to get a P570WM with 780s and an 4930K.,,,,

 

It's a shame to have to spend extra money to get what you paid for to begin with, but if I were in your shoes I would be paying that 15% fee and moving on. I think 880M might end up working right with more vBIOS tweaks and driver optimization. If that happens I might upgrade later on, but I'm glad I didn't jump on the bandwagon as an early adopter. I suppose whether it ends up being fixable will depend on the root cause.

I am still struggling with the temptation to buy a P570WM. I don't actually need one, but I want one for number chasing. I could get a barebones model fairly inexpensively and put some of my spare parts back into service. I wish the M18xR2 motherboard would support 4930K CPUs.

 

@Mr Fox:
So let me quote your exact post:

My reply was:

So now you are saying that the 880M suddenly is working like it should? After I pointed out that the CPU performance between yours and Ethram is miles apart...? Dont go backstepping now Fox. Thats what you said above. 880M is 2.5% from the 780M. You said you ran the 780M on the same clock as 880M yet say "one should expect more"....How does that work?

Later you said

No it didnt. You scored less than 1% more.

You even write this in the link description

Yet that particular comparison shows the two GPUs perfoming the same but the CPU doesnt.

And you follow with

GTX 880M is a 780M. Why do you expect more? If you clock the two the same, 880M shouldnt magically score more because it features a different name.


I can`t find a single post in this very thread that shows 880M (the GPU, not the system) performing less than it should (except the driver flag thingy).
Not hating on your party here, but it all seems very weird and funny at the same time

 

If you and Mr Fox ran the CPUs at the same clocks, there could be something wrong with your CPU yes.
He score 10299 in 3DM11 in CPU score, you did 8097.

 

If you paid for the best and owned the "newer" product would you be satisfied with a 2.5% difference in favor of the older GPU? It's not a mystery to anyone that 880M is a 780M with more vRAM... well, at least in concept. There is something different going on that is causing it to not live up to expectations. This is the part of the discussion you are missing.

Yes, close but no cigar is accurate. What you apparently are not aware of is the fact that the system I used for comparison is not able to consistently produce results that are within 2.5% of 780M SLI running at matching clock speeds. Those are the best examples on the days it worked right for Brother Ethrem. Picking apart my posts bit by bit is not going to convince anyone that you're an authority on the subject. You are trying to use logic (which should work) in a situation that is not logical. Something is wrong with 880M. Either bad batches of hardware, something in drivers or vBIOS because most of them do not work right. We don't know exactly what the problem is right now, only that there is one.

 

I think what Cloudfire is trying to get across is that when a 780M is clocked to the same speed as an 880M, you shouldn't expect the 880M to perform any better because the 880M is simply a higher clocked 780M with more vRAM, and it being 2.5% better is already a bonus.

Basically he's saying the 880M uses the exact same chip as the 780M except clocked higher (so probably binned higher). So when the 780M matches clocks with the 880M, it essentially becomes an 880M.

 

But that's just it. The 880m doesn't perform at stock speeds all the time. If it's driver crashes 4-5 times. That's it, it's game over unless you emulate a 780m. Another problem is the latest driver - 340.43. It's down clocking our 880m's to a horrible; 130mhz at times. Not to menthion, it cripples the modified vbios.

 

I realize what he is saying and it's not wrong conceptually. It would be accurate if everyone had a properly functioning 880M powered system. It also has nothing to do with the context of this thread. He is insinuating nothing is wrong with 880M and everyone that owns a system that malfunctions is somehow mistaken "and barking up the wrong tree" for thinking it's broken. It clearly has performance and instability issues.

The benchmark comparisons were not offered as evidence it is messed up. I was excited to see a rare example of an 880M powered system that was having a good day and working better than most of them do. At that point it was the only example I had seen. I took the opportunity to run some benchmark comparisons to show it was keeping up with my 780M system. It wasn't keeping up, but it was very close. Ignoring the Haswell performance issue that affected the combined score and physics score, the actual graphics score was decent... 18196 for 780M versus 17684 for 880M... 2.9% difference there. And, the CPU does not affect the graphics score enough to make a difference with 3DMark11. That's what makes 3DMark11 one of the best tests for measuring the graphics score. You can run tests 1 through 4 at 2.9GHz and get essentially the same physics score as you do with the CPU running at 4.5GHz.

That's a small difference, but if I had just dropped a ton on money on today's best that would be a disappointment. I could live with identical performance and twice the vRAM as an added bonus because 780M rocks and at this point in time there is nothing better. 880M doesn't necessarily need to be a lot more powerful to be awesome, but a performance decrease is not something to be excited about. That 3% difference in favor of 780M pales in comparison to the instability and lackluster performance issues most people are having with 880M. It's messed up and the benchmark comparisons I provided do not prove it is functioning properly. They only show what it is capable of at times when it does work correctly.

Exactly... there is a problem, but nobody needs to tell you that because you're living it.

To correct one of my earlier comments that Ethrem had the only example of a properly working 880M system, you did too... for a bit. The 880M seems to be a moving target with inconsistent results from one day to the next, or one driver to the next. We haven't seen this kind of inconsistency to such a grossly exaggerated extent before.

 

what we need is a new semiconductor company that which can innovate and create new products. Since its dropped down to nVidia and AMD, its been stale. When it was a close call for either side, things got done. But now, its as if they work together to maximise profits and release products together without trying to out due the other or gain any advantage over the other by new technology. I say this, because theres been nothing new for.... wait, let me think, when did the GK104 chips come out, like 5 years ago right?

 

I think the new consoles have alot to do with it too. If not directly then indirectly. Since they are grossly underpowered and most games are multiplatforms the Devs work within a certain range that doesn't create enough lust for more powerful PC hardware.

 

Well, they shouldn't be allowed to get away with advertising this way if it's not true. They need to fix it or change the description to make their customers aware that this an expensive and powerful GPU that malfunctions.

 

Unfortunately I believe you are right in this way of thinking and I have been having the same thoughts myself recently as well. All you have to do is look at Watch Dogs to draw that conclusion.

On the point of the 880M, admittedly my system has not gotten unstable on me once since I flashed back to the stock vbios on 337.88, even with a temporary flash to the mod in the middle (which was previously unstable period). I had Bioshock turn black and freeze the system once but other than that, it's been totally stable even in the train wreck that is Watch Dogs.

It's exactly as Mr. Fox says though - the cards are unpredictable at best. At the same time though, I can't remember a new graphics card that ever was stable at launch, even a refresh, so I still have hope that future driver updates can fix this card.

I will say that it performs flawlessly in Linux so that gives me great hope that it's just a matter of driver optimization.

 

It says a lot when you can match my boost clock speed at 993mhz, and yet still beat my score in 3D Mark 11. I'm talking about the graphics score alone now, not the combined score. It's a real shame. I really hope Nvidia come through for us on this one.

 

@woodzstack: You posted this over on the nVidia forums, so I'll just quote it here for everyone's convenience:

If it would really "only" cost a few hundred dollars, I honestly wouldn't mind pitching in say $50 if it means we could actually force nVidia to pay attention by costing them business. If we get enough people pledging to pitch in here I'll put down $50.

 

Legal pressure has not worked in the past with NVIDIA. Some companies go into "war mode" when you do that, so there is a danger that nothing would get fixed. I have worked for a couple of large corporations that take this approach to legal challenges. The idea is to kill the tail that is trying to wag the dog, and it is better to spend millions of dollars to win the fight than to admit to making a mistake.

I was part of the last Class Action lawsuit that dragged on for years. NVIDIA never admitted to anything and the solution the court approved was to have my $900 HP laptop with nice features replaced by a totally stripped-down $299 Compaq with a pathetic AMD integrated GPU offering half the performance and hardly any nice features. Score for NVIDIA 1, consumers 0... after several years of the HP being a dead and unusable door stop.

880M is brand new and I think it is premature to try using force. Better to make a big deal about the problem and give them time to fix it through driver or vBIOS updates. That will let them act as though nothing was wrong instead of having to take a defensive. To put things into perspective, 7970M owners put up with all kinds of problems and waited more than a year for decent drivers to be released. 880M has been out for what now, maybe 60 days?

 

Enduro was to the 7970m as what Madoff was to corporate America.

 

Fair enough, hopefully our patience won't be in vain.

 

Looks like you've already had positive influence on a poster in the GeForce forums Mr. Fox. That's probably the best course of action there. Well done!

 

I hope that turns out well for the guy. The best way to get NVIDIA's attention, I think, is through the OEMs. If you stop and think about the situation, end users are not NVIDIA customers... at least not directly. Nobody buys video cards from NVIDIA. OEMs like Dell/Alienware, Clevo, Asus, MSI, EVGA, Gigabyte, etc. (and probably a few manufacturing sub-contractors with names we have never heard of before) are NVIDIA's customers. NVIDIA licenses their technology to OEMs that produce products (or sub-contract with other vendors to produce them for their brand) and the OEMs' customers are end users. If a substantial number of people that purchased systems with 880M start asking to return them it is going to create a ruckus behind the scenes. Even if some of the OEMs refuse to allow RMA they will notice and go back to NVIDIA to complain about the bad experience it is causing for them and their customers.

Some of them (including Alienware) might even go as far as placing an "engineering hold" on distribution and not offer them as a configuration option if they start getting a number return requests from dissatisfied customers. This would be more likely get NVIDIA's attention. They are more likely respond to something that affects cash flow than legal sword-rattling.

A potential down-side to having product distribution placed on hold is that they may be unable to offer previous generation hardware options, meaning they may not be able to offer 780M instead of 880M. If you're ready to buy and the only options are lesser performance video cards, that's going to make some shoppers very angry.

 

go to there face book page and spam the link to the nvidia forums and demand an answer! if we can enough people on it and get Reddit and 4chan lol on it maybe we can get some where..

 

Still not a peep out of NVIDIA, but a couple of morons showed up, LOL.

Maybe they'll just quietly fix it with a driver update and pretend it was never broken. If that really fixes it, I guess it will be OK. The fact that they are not touching that thread with a 10 foot pole tells me they know they screwed up. Otherwise, they would be in there asking questions to gather more information.

 

definitely a public relations type thing

 

Well I've replied in there official 340.43 thread as well as submitting feedback in the link provided by the moderator in that official thread. If we don't hear anything soon, we may as well just give up hope with the 880m guys.

 

This whole thing about the 880M being a pile of dog meowcow is annoying me. nVidia has been the company I tell people to get if they DON'T want to deal with problems with drivers and hardware. Even if it's just one card, I'm feeling utterly awful about the whole endeavour. I'm having serious issues right now with my P370SM3, down to needing a board replacement, and I swear I've been so scared that if I needed a new machine I'd *have* to get 880Ms.

Also, I've noticed I've migrated to the alienware forums. I must admit, while I'll likely never own an alienware as long as I live in this country (because alienware = steal bait here, while my P370SM3 looks like an old business laptop to most people) and/or have to somewhat budget with my high-end laptops, I do like the community you guys have. Mr. Fox is especially aggressively helpful XD.

 

If the board replacement is for P370SM3 then you'll get to keep your 780Ms, since the P370SM3 doesn't officially support 880M anyway unless you flash Prema's BIOS.

Yeah the Clevo forums doesn't seem to be getting a lot of traffic these days, but the AW forums is always teeming with activity.

 

A lot of people posting in the Clevo forum now are former Alienware owners, such as myself too.

My cards stopped misbehaving with an OS reinstall. They still throttle but they're still faster than 780Ms at stock clocks and do overclock quite well, I just can't get rid of the heat. The highest overclock I got before I overloaded my adapter (both of them actually) was 1153/1475 @ 1.1v but the cards went over 90C just running Firestrike, 3DMark 11 overloaded the adapter, and Catzilla throttled at the plasma/smoke part of the test (I forget what its called right now) due to both cards hitting 93C.

Here's the Firestrike from that: NVIDIA GeForce GTX 880M video card benchmark result - Intel(R) Core(TM) i7-4940MX CPU @ 3.10GHz,Notebook P377SM-A

Since they were both at 1.1v, I think I was pretty close to the limit for these cards anyway but that's not a terrible overclock either. That Firestrike is extremely misleading because I knew I was approaching the limits of the A/C adapter and so I had turbo off for that test but its proof that the card at least hit those clocks and was stable enough to finish the test.

 

I like how all the morons in that thread thus far have all been desktop users, and the few intelligent ones all have a laptop. Hmmm...

 

There's rumours that we may see a new driver this week at some point - 340.51. Hopefully this turns out true, so we can resolve the 880m problems.

 

or it could make it worse

 

Looks like a call to action... Click Here to Act if you own GTX 880M and want to see it fixed.

 

Brilliant if this issue gets addressed for the users out there who care about getting what they paid for. Although too little too late nvidia. I won't be moving to 880m anymore, nor will I jump on and get bent over on your next offering, not until its been closely scrutinized.

 

I went and added my info.

 

I've supplied my sky diver results in the official beta driver thread. I've also given a full run down on my problems with the 880m. Manuel has also seen my results, so we should see results soon.

 

I decided not to go the benchmark route and instead posted my Bioshock Infinite time stretch showing an actual game throttling the cards below their stock speeds. I'll have to go over and see what they're saying about benchmarks though.

 

Amen. I'm actually terrified that Maxwell will simply be "efficient" and nothing more. Though with the leak of the 375W desktop 880 prototype I'm hopeful performance won't be sacrificed in the name of efficiency.

 

The efficiency could bode well for us even if it's only a slight increase in actual power. The efficiency could mean lower temps, better OC-ing headroom and in general better cards. But that being said, that's all conjecture. The 860M is a great example of a midrange maxwell card though, compared to the previous 560M, 660M and 765M cards, if I remember correctly.

 

Here's my post in the official beta thread - GeForce 340.43 official thread

It appears that driver 340.51 is nothing than Nvidia VGPU software - Link

 

As a new owner of an Alienware 17 with a 880m GTX, I find it hugely disappointing to see a supposedly flagship mobile GPU SHACKLED like this. Nightdex I read a post of yours mentioning you were using a modded vBios. Does it help? I don't really plan to OC my card but I expect to get a least STOCK speeds when gaming...

 

It helps improve things significantly. You'll notice straight off the bat that your 880m will run at its boost speed of 993mhz. It also benches well with the unlocked vbios.

 

Just watch your temperatures. 993MHz running constantly pumps some serious heat.

Sent from my HTC One_M8 using Tapatalk

 

Not for me. 4 hours playing Murder Soul Suspect pumped my temps up to 81c with Liquid Ultra. 1051mhz bumps me up to 84c. Any higher than 1051mhz, makes my temps rise to 90c flat after 1 hours worth of gaming.

 

Most of us don't have liquid ultra. I think it's safe to assume that specific poster doesn't since he said that he doesn't plan to overclock.

Sent from my HTC One_M8 using Tapatalk

 

That poster isn't taking any chances!

I live in Dubai. It's midnight and it's 36 degrees celcius. Max temp today was of 45 degrees... I am going to keep things COOL for the time being

 

My temps weren't all that bad with IC Diamond 7. Liquid Ultra doesn't have that much of a huge impact on my temps.

 

Stock I average 82C with an average clock rate of 926MHz on my secondary card with automatic fan control but modded vbios puts that card to 93C in 20 minutes running 993MHz with the fans on max. That's with IC Diamond.

My primary hits 91C.

The cooling on the primary is much better than the secondary and it still skyrockets. Your liquid ultra is probably doing more than you believe. At the very least it is delaying how quickly your temps rise.

Sent from my HTC One_M8 using Tapatalk

 

You also have to account for the cooling Alienware offer.

 

Ethrem is right... Liquid Ultra makes a huge difference, and perhaps more than you might recognize. If you're not pushing a CPU overclock the benefit will not be as obvious with an effective cooling system like our Alienware systems have. However, with higher clock speeds and as more heat is generated, the benefit becomes far more apparent. I can just barely manage 4930MX temps at 4.3GHz with IC Diamond. It's OK with the fans running full blast, but hotter than I would like it to be (90-95°C under 100% load, compared to 80-85°C under 100% load). With Liquid Ultra I can easily run 4.5GHz-4.6GHz with slightly better temps than 4.3GHz using IC Diamond or NT-H1.

 

Hey guys... looks like Manuel is looking for more feedback. They want system BIOS info as well.

 

This is good news. Wheels are in motion. Isn't it good when companies listen to their bread and butter.