Just got my 880M twins!

Not really that much abuse... I think most people don't really know what their systems are capable of and assume what the Alienware has provided is ample. But, it's not ample. GTX 780M SLI on the Alienware 18 with NO GPU overclock pulls 335W in 3DMark11. In this video and you'll see 550W+ from 780M SLI without a lot of effort.

<iframe width='640' height="360" src="//www.youtube.com/embed/7CoeOq-lkdY?rel=0" frameborder='0' allowfullscreen=""></iframe>

 

Wow that's pretty nuts that they cut it so close. If that laptop was a subwoofer though, that'd be a perfect match

 

Yep, that's cutting it really fine with a 330W adapter & stock 780M sli, although your CPU was at 4.3Ghz there, that's a pretty good overclock isn't it? I guesstimate your CPU is using about 90W during that test? (And your measurement was probably at the 'combined' test on 3DMark11 - close to 100% CPU & GPU usage). I would guesstimate that with a standard 45W i7 CPU (4700MQ), then this would give you an extra 50W of headroom, which would allow pretty good gaming overclocking on the GPUs to stick within the 330W adapter limits. Am I far wrong thinking this? Your CPU is taking up a big chunk of that wattage isn't it?

 

Well, no... 4.3GHz is not spectacular... at least not in my mind. I'd consider it a nice everyday "average" overclock and nothing worth bragging about. Both the M18xR2 and 18 I have here run 4.3GHz as their normal everyday clock speed. I do literally everything on both machines at that clock speed... even just web browsing and email.

Your thinking on power consumption is fairly close, but the rationale is flawed. Not on your part, but unacceptably flawed on the part of the manufacturer. I really don't care that downgrading to a comparatively incompetent CPU like the 4700MQ gives more headroom than a 4930MX. If they have no intention of supporting the 4930MX, it should not be offered as an option and they should stop defrauding customers by accepting their money for an unsupported hardware configuration. The rationale is also flawed because you can cripple the 4930MX to mimic a 4700MQ and the system still cannot power 780M SLI to the limits of its functionality.

Anything that can be placed into the category of "pretty good" sucks if you care about getting what you paid for... a machine that works right. If you can exhaust the AC adapter running the CPU and GPUs within their functional limits, or if the machine turns itself off when pressed to a certain load point below its functional limitations, then the design of the machine is inadequate, fundamentally flawed, and anyone that buys it is being cheated. To put it bluntly, for something as petty as the capacity of an AC adapter or an arbitrary limit similar to the capacity of the AC adapter to ever become an impediment to operating a system within the confines of the functional capacity of its components is nothing short of retarded. It also smells dishonest to sell something that is gimped.

 

Sure, if they're selling the 4930MX, which is an enthusiast overclocking CPU then it needs the headroom for overclocking, otherwise they're selling you short - I definitely agree with you on that, the power adapter is not enough for the system, a 4700MQ allows a little more overclocking on the GPU's, but as you say it's a compromise that people who buy this machine shouldn't have to make. MSI have also routinely been doing this (180W adapter), but worse, with there NOS battery system - which is sold to consumers as a plus that it needs to take battery power plus adapter power to sufficiently power the laptop! At least Alienware haven't resorted to that kind of rubbish. I don't know why manufacturers skimp on the power supply, I mean how much money are they gonna save doing that versus the amount of pissed off owners with gimped laptops!

 

You know all this talk of 4700MQ just made me think of something: it seems possible that Alienware designed their machines around the 4700MQ, and then realized aftewards "oops what about the 4930MX that already draws 10W more at stock? Oh well we'll just lock down the BIOS to limit power draw. Hey look problem solved!"

 

I think it's based around money and/or reliability worries, which all comes back to money in the end. But selling customers short will bite them in the *ss eventually!

 

Anything is possible. I think it probably boils down to either not paying attention (poor quality control) or an intentional show of poor judgment with a decision to cripple their machines under a flawed concept of reducing warranty claims. Or, maybe something we haven't even considered because we're jaded on the receiving end of mediocrity. No matter the cause, the end result is damage to the brand reputation and alienation (pun intended) of their customers.

 

So about 270-300W actual consumption from the machine in 11 with no voltage mod, shows the brick is pretty close to maxed with dual cards.

 

I remember in 2006 trying to overclock my Geforce Go 7600, 9 out of 10 replies on any forum about overclocking my laptop were met with scoffs and patronizing remarks. It's awesome that a genuine laptop overclocking community has grown, judging by Mr. Fox's posts.

 

Actually, I think the system was designed the logic of a 4930MX costs X amount to replace, so if we limit what end users can do to mess it up we can potentially save X amount in support and warranty costs.

 

Could definitely be that... although, I'd be curious what that kind of examples such paranoia would be based on. I'm not saying it doesn't or hasn't happened once in a great while because I have no doubt that it could under the right set of conditions. But, I have not seen a mobile Extreme CPU damaged from overclocking before and don't remember seeing an Alienware owner post about that before. I've seen a few that began life already messed up and needing to be replaced under warranty due to a material defect, but not a good one that someone actually messed up from overclocking inexperience or frequent benching, etc. Some people can break anything they touch, but managing the masses based on the lowest common denominator or what could happen as an anomaly is not a sound approach. So, I hope that's not what is driving it because it's a serious flawed perspective if it is.

 

Definitely a contributing factor.

What I struggle to understand is this: even though the BIOS is locked down, you can still access most of the settings via XTU/ThrottleStop. Dell must know this (or at least I hope the engineers are aware of this), so really what is the point of a locked BIOS in this case, to make life harder so we would feel discouraged? I mean if the lock could be easily circumvented via a freely available software, that lock is pretty much useless.

Yeah I understand there's loads of other locked stuff that can't be accessed via XTU/TS, and the BIOS cripples OC performance. But overclocking is still possible. If Dell truly was trying to prevent us from doing that, well then they have failed.

 

I think laptop overclocking is awesome, often bringing more gains than overclocking desktops. I've been lucky in accidentally choosing chips with good overclocking headroom. I had a Dell M1530 before this laptop with an 8600M GT chip (bought in 2008). Stock it was: 475/950/700 (core/shader/memory), but I managed to overclock it on stock voltage to 710/1700/1000, which is a 50% overclock on the core and a 79% overclock on the shader (43% overclock on the memory), a big effect on gaming! That laptop never died on me even though it was overclocked the whole time, used it for 5 years. This 670MX also overclocks like a beast too, with an 87% overclock on the core (added voltage), also as a gaming overclock. For my next purchase I imagine I'd also choose a good priced GPU with excellent overclocking ability (if they exist in the future!).

 

Wish my M18x R2 would just work....

 

What happened?

 

Nothing, it never worked after I upgraded it to 880M's.

 

Brother Fox,

It is not so much the cost of replacing an extreme CPU if it is damaged. It is more or less the cost to replace the entire system if something else is wrong. That cost for a new system exchange with an extreme CPU and maxed out spec's cost's Dell a ton. That being said, Overclocking can damage other components as well, such as the MB, and RAM, which during the troubleshooting process will be changed out. We all know CPU's rarely die, it is everything else that the CPU interacts with, and the eventual cost of a possible system exchange that is driving Dell\ Alienware to back down overclocking options. At least that is my theory, I could be wrong, but I am pretty sure I am right that there is a substantial cost to the company when something breaks and it is due to improper overclocking.

 

Aritoshed, is your stock GPU's not working now?

 

Your theory could be right, especially if they are suffering from paranoia over what might happen. We see examples of tons of system exchanges that don't seem to have any relationship to overclocking. If overclockers are such an insignificant minority as some people like to say we are, system exchanges that were necessary to do improper overclocking would be a commensurately small number of the total system exchanges or major part replacements. I think in many ways some people are their own worst enemy. I suspect a lot of system exchanges are totally unnecessary and offered too eagerly in an effort to appease whiners and unreasonable people that think they are entitled to a system exchange as part of a 3-strikes rule. In other cases they throw parts at a problem a few times, sometimes the same parts more than once, and then decide to replace a system because they have no idea what is wrong with it. That may actually save them money compared to continuing the exercise of multiple part dispatches. Every time they have to open a support ticket and send parts, dispatch a tech, or initiate a depot repair it costs a fair chunk of change.

Regardless of what is driving it, I tend to view having the new systems locked down as being an unethical demonstration of control. I recently saw a response from a respected Dell employee in another forum, both of which will remain unnamed, that really struck a nerve with me. The response was to a customer complaining about his new Alienware beast being crippled. The employee's comment was a canned response (probably drafted by a member of their legal team) which reflected a mentality that the Alienware system in question was their machine and they have reserved a right to control what customers are allowed to do with the products purchased from Dell/Alienware. I was floored by the comment and haven't really recovered from the shock of that expression. I may never recover from it.

That is a good question. If the machine doesn't function exactly the same as before with the original graphics cards installed, there could be something else going on. The 880M cards might be bad, and they may have caused damage to the motherboard if the original parts no longer work right.

 

nevermind.

 

But dell/alienware also don't help themselves when the manufactures send out wrong parts. Like take me for an example. I called many so called tech specialist from Alienware. Not one of them could give me a solution to my 880m down clocking. They send an engineer out only to tell me that they sent a 180 watt power supply for a GPU that demands almost all of the 240 watt power supply power. Makes sense right? No didn't think so. Anyhow, when there own engineers tell you that there "tech specialists" are daily simple minded and not technical minded in the slightest. It makes you wonder why Alienware or Dell still exist as one of the top leading laptop brands. I expect to experience a present straight forward customer service . Not customer service experience similar to what Acer opt for. So going on what Wheth stated about Dell worrying about the cost of replacements. I doubt that they care since they can't even gather the correct parts that have been ordered for a particular configuration.

I'm with you brother Fox as well as John. Alienware are crippling there core audience, for what reason I'll never understand.

Yes guys, I own the 17 and 18. I don't have a problem with Alienware's build quality as it's the best out there for a laptop. But I also expect to configure my highly priced Alienwares as I see fit.

Yeah you may be correct there brother Fox. It's seems very odd that Aritoshed's stock GPU's no longer function. I would place my money on either a dead board or like you mentioned, thise 880m's have killed the main board.

 

Yes, couldn't be happier with the build quality for the M18xR2 and the new 18. And, I really love how easy the new 18 is to work on. I actually enjoy taking it apart because it's so nicely built. It's just a shame it doesn't perform at the same or better level as the M18xR2. I'd forgive the missing features on the new machine without much thought if it was just as potent, could use the dual AC adapter and the BIOS were not locked down.

 

Brother Fox,

I couldn't agree more. As to what that unnamed Dell employee said, that is actually the sad reality of society today. The control over their product is legally acceptable, and is the equivalent of the product use warning on aerosol cans. It is their way of legally defining the way their product must be used in a safe and responsible way that way if someone abuses their computer they can deny warranty, or they establish a standard that can be used in legal cases where they might be liable if someone is claiming their computer burned down their house. I think it is just a byproduct of our society, and I wouldn't go as far as to say Dell is trying to take control away from the consumer. I think it is just Dell protecting themselves, and I can only hope that reasonable judgement is used generally when dealing with consumers.

 

I hate it when companies sell a product and then still consider it theirs.

 

. Just got a call from Dell telling that my computer will be 8 weeks delayed due to part delivery problem.

So, cancelled that order...... So now I have 6500 dollars to use on a laptop but no laptop to buy.

 

It's a hard life! ;-)

 

When is your P570WM supposed to be delivered?

Just an FYI...

 

1378 points? thats low?

My 570M was supposed to br delivered in the middle of July.

 

What is going on with these cards?

 

I wouldn't mind a donation

 

Dunno, seems to work under benchmark but after 5-10 seconds the FPS drops like 90%, temps are around 70 degree. Both cards are also showing full usage during benchmark.

 

I just did a Alienware diagnostic run and both the videocards, CPU and RAM passed.

 

Yeah, that is absolutely horrible. And, it is not off by just a little bit... it should be much, much higher. The GPU score should be about 8 times higher than what it is in that example.

Good question. It's comforting to see the same problem with Clevo in kind of a selfishly sick way, but nobody deserves to have something perform that bad. I haven't see a single example of good results with 880M except for the engineering samples that John and svl7 had the opportunity to tinker with before they were released through normal channels. Not saying there are no good examples anywhere... only that I haven't seen one yet.

 

NVIDIA GeForce GTX 880M video card benchmark result - Intel Core i7-3940XM,Alienware M18xR2

 

That's really sad, but not as sad as the 3DMark11 score from that Clevo owner.

There has to be a solution to this. Not sure what it is right now. Something is really wrong.

 

When I did the 3Dmark11 test, the FPS was high but it drops many many times during the test lowering the avarage score..

 

do these cards at least work fine stock, in the laptops they come in ? like the AW 17... ?

Oh and fox, it seems the guys who tested the first few runs of those 880m's must have just been given deliberately special cards or they were complete enthusiasts hippies about the cards, or something.

 

B

Nah, I have them. They don't perform as well as I hoped but I'm not having the performance throttling issues that others are having.

Nvidia really pushed these cards. I've never owned a video card before that couldn't be stable at stock clocks with a slight undervolt yet these cards I have will crash with a 7.5 UV and even with the molded vbios, I can squeeze out no overclock on the stock voltage, something else that I have never seen before. I can get 450 on the memory but I can't get anything on the core.

The bottom line is that nVidia shouldn't have released these cards. They have officially pushed Kepler to the breaking point and I have zero faith in these cards. Even when they are performing well they will suddenly stutter while showing no drop in the core or the memory clock. I think these cards are going to have a very high failure rate. I hate to say that but when cards are this unstable, it means they screwed up the design. The last time I saw such a problematic card was the 7900GT. That also was the first and only card to burn up on me with stock clocks.

I have a small amount of hope that driver maturity could bring out better performance and overall stability but these things are really being pushed. It is possible that in my case a voltage bump could help but I'm already pretty high on my thermals so I haven't explored that option yet.

Sent from my HTC One_M8 using Tapatalk

 

Don't think drivers will help, as you know they're just Kepler cards, which have been out for about 2 years already. Could the stuttering you have be down to sli issues, especially seeing as you say the core clocks & memory clocks don't throttle when you see these stutters?

I do agree with you though, they've pushed it closer to the limits of stability in terms of clocks & voltage with the 880M's, they allowed more headroom with the 780M & 680M. Also closer to the thermal limits too. They should just go ahead and release Maxwell high end already!

 

Actually, that is not accurate. The stock clock speed of 880M is nowhere even remotely close to maxing out the limits of stability for this GPU... it's somewhere about 300MHz short of that danger zone. Worse case scenario we would see elevated temps compared to 680M and 780M at the 880M's minor bump in stock core clock speed. The problem is something else, not clock speeds. What we see happening proves that 880M is not simply a rebranded 780M with more vRAM. It's Kepler, yes, but they broke something real bad that wasn't broken this way with other Kepler cards. Something with the vBIOS or hardware architecture is borked, and potentially exacerbated by the drivers.

 

Unfortunately it happens with games that aren't SLI enabled and with SLI disabled as well. I don't know what to make of it, I really think it is a voltage limit though. If I force them to 970 instead of 993, it stops happening and its smooth sailing.

If it wasn't for their hitting almost 90C under load (one at 89C, the other at 87C), I would tinker with the voltage a bit.

I've checked my pads and everything that needs to be covered is covered so I don't know what to make of it.

As it is, for fear of long term reliability with these cards, I'm going back to the stock vbios. The last thing I would need is to have one fail with a modded vbios voiding my warranty.

The performance of these cards is disappointing compared to the other Kepler cards but it's out of the box performance is still top tier. Nvidia should feel embarrassed though that's for sure. They pulled an AMD...

Sent from my HTC One_M8 using Tapatalk

 

You're right, it certainly could be a power problem, like voltage being too low or something like that. It could also be a power draw problem for some folks. I'm not very confident all of the laptops with a 880M stuffed into the PCIe slot are up to the task of powering the sucker properly.

 

I'm really happy to hear you say this. I was somewhat under the impression that Nvidia simply made an overclocked GTX 780M and called it an 880.

 

What's not accurate? The 880M is operating closer to the limits than the 680M & 780M. You know that the 880M is an overclocked 780M, so it has to be operating closer to it's limit at stock, as Ethrem explained that he has no scope at all for core overclock at stock voltage and zero scope for undervolting at stock clocks. Also, when you say the 880M isn't close to it's limits, it is in the majority of notebooks out there, from a cooling perspective. And I have a feeling the limits that you talk of are benching overclocks, not gaming overclocks. You said the 880M was 300Mhz short of the danger zone, I guess you mean for benching, I don't think I've ever heard of people gaming with 1254Mhz core clock speed on a 780M or 880M, which would mean a core voltage of what...nearly 1.2V!

 

Maybe I misunderstood the implication was merely "closer to" and not "close to" the limits. There is a subtle difference there, but the meaning is different. Sure, it's possibly closer, unless the limit was raised... then maybe it is not closer to its function limit than 780M was to its limit. We don't know this and I think that statement is a wild guess. Nothing wrong with wild guesses, especially since that's all we have to work with at the moment.

I still believe the issues people are facing have nothing to do with the core or memory clock speeds being higher than 780M. It's easy to say an 880M is an overclocked 780M. We all know it's similar, but the behavior is not that of an overclocked 780M even by a long shot. Something is different here... something is broken that has not been fixed yet. Hopefully, it will be fixable.

 

Are you recommending not to buy any laptop with the 880m GTX cards ?

 

Kind of hard to recommend or discourage the purchase of something I haven't tested myself, but I know I would be very apprehensive buying something with 880M based on the chaos around how crappy it seems to performs for most people. Every new Clevo and Alienware example I have seen with 880M doesn't function very well. I would want to see more examples of it performing as well or better than 780M. However, the GTX 780M wasn't a great performer at first either... I'm taking a wait and see what happens approach for now. At this point I am kind of glad I did not jump onto the upgrade bandwagon. The current situation could easily change for the better with driver optimization and maybe another vBIOS tweak or two.

 

Is there anything that gtx880m has 8 gb memory and not gtx 780m? Have they made ​​any other changes on this card is not good?

 

I just sunk $3500 into a brand new gaming laptop (was rocking a GTS 360M for something like 4 years now), so I'm biased to any information that says the 880M has some secret, hidden, heretofore unknown benefits over a GTX 780M.

But the information is conflicting. The manufactured card is the same as a GTX 780M, but at the same time, mentions have been made to other differences. The Techinferno review mentioned improved ASIC quality for example, leading me to wonder what other differences there might be.

Mr. Fox seems to be referring to a negative difference, something hindering the card. I continue to hold out hope that the performance is there to properly dominate it's predecessor, as it seems Mr. Fox is inclined to believe, it just needs to have it's potential unlocked.