Alienware M17x R3: Random screen blank + system hang

This is a 64bit system with Windows 7 SP1. I'm running with the latest NVidia drivers and the latest BIOS (A12) and Video Bios (70.24.2B.00.06, A01)
RAM: 16GB
HDD: 250GB SSD + 750GB SATA
Graphics: Nvidia 580M with 2GB VRAM (no onboard card)
Screen: 120Hz 3D

Problem Description:
This problem started a couple of months ago where my screen would go blank and the machine would need to be shut down hard. The keyboard lights would stay on, as would the fans, so it's not like the entire system powers down. Pressing caps lock or any other keys doesn't do anything so it's not only the screen that blanks out (backlight is off as well), the system does hang. There is no pattern to this, it could even happen pre-boot or after I have been using the machine for hours (details below).

It happened infrequently at first, so I didn't pay much attention. The problem has progressively worsened to the point that I sometimes cannot even get the machine to start completely for hours. At the same time, sometimes it could work just fine for days at a stretch without exhibiting any the problem.

Symptoms:

The machine fails in a number of different ways:
A) Before OS
1) Start machine, hear 8 beeps repeatedly (rare).
2) Start machine, hear 6 beeps repeatedly (very rare)
In most cases, subsequent starts do not sound any beeps.
3) Screen is blank on start - don't even get POST screen. The machine remains on though, but doesn't appear to be proceeding with the boot. No beeps though.
4) POST screen appears for a few seconds, screen goes off almost immediately afterwards. Then same as #3.
#3 and #4 could happen multiple times but after several attempts the machine would suddenly work (or exhibit other symptoms)
5) While running the Dell diagnostics (from the F12 option) - screen would suddenly go blank when running the *thorough* memory or video card tests. I suspect the problem is not with the part that is being tested, but again the intermittent nature of the problem with the machine being on for a longer amount of time.

Given that these problems are manifesting themselves before the OS is loaded, I don't think we have an OS problem here.

B) OS Related
6) At the "Windows did not start properly, what do you want to do" option, I press a key to prevent Windows from starting with a default. When left in this state for a few minutes, screen could go blank, system has to be hard shutdown.
7) Just before the switch to the Windows is starting to the login screen, screen goes blank, system must be hard shut down. This is the most frequent manifestation of the problem.
8) I am able to login - Desktop appears, screen goes blank in a few seconds or minutes - before I can actually do anything.
9) Same as #7 but here, even though the screen is blank, windows continues to start. I hear the login sound, I can type in my password (blind) and windows starts. I can the run programs and even shut down cleanly (all blind, using keyboard shortcuts). If an external monitor is connected, it does not help, both screens remain blank.
10) System works fine for an entire day (problem might occur a couple of times, but a restart keeps it working for hours), am able to shut down at the end of the day.
11) Problem will manifest itself in safe mode as well.

I have had the the graphics card head cleaned and heat paste reapplied for both the graphics card and the motherboard. The fans appear to be functioning fine and if I keep a temperature monitor open, neither CPU nor GPU appear to be overheating (unless something causes the temperature to suddenly spike to some very high value in a second before causing the system to crash). Even when gaming, the graphics card appears to be around 70 degrees centigrade.

My best guess now is that the graphics card is bad, (I wish this system had an onboard card too) but I cannot explain the intermittent behavior at all. The suggested next step from Dell was - replace motherboard, graphics card and heat sink assembly and that seems rather excessive for me, if I were to do this, it is very difficult for me to even get an estimated cost out of Dell. I'm not sure if they can replace just one part and let me have it for a week to see that the problem does not recur else given the intermittent nature of the problem, the replacements might work for a short while and then the problem could manifest itself again.

My original thread at Dell is here: Alienware m17x random hangs - screen off, system on - Alienware Forum - Alienware Club - Dell Community

Sadly I'm out of warranty; being stuck with an unreliable $3.5K machine after just 2.5 years of use is extremely disappointing and frustrating, other than this the machine looks like it can hold its own for at least another couple of years.

Any suggestions much appreciated.

 

Hi there. The first lines tend to say it all .

The 580m has been a notoriously bad card with multiple failures on machines. It seems that now they are getting to an age where even the good ones are failing. I see at least one a week currently!

Dell's answer to throw everything at it is generally what they would do under warranty.

The chances that it is just the NV GPU is high. It may have been overheating and you didn't know?

You might find driver version 314.22 ( NVIDIA DRIVERS 314.22WHQL) gets you some more life, it has worked for a few, but basically you need a new card. Had exactly the same happen to one of my (580m) machines, luckily in-warranty.

Do you mind loosing 3D capability?

Try a search for 580m failure - you should come up with quite a few - or check out my posts

 

From this description, I can think of three possible issues: RAM, GPU, or motherboard.

I suggest taking out one memory stick at a time and restarting to see if the issues (or some issues) go away. If this doesn't work at all, then the issue is most likely the GPU. Given the fact that you're not experiencing BSOD's, I doubt it's the motherboard. As Micky said, the 580M is an unreliable GPU.

Your best option is to find a GTX 680M on eBay and install it. The 680M is much faster than the 580M - 580M SLI performs about the same as a single 680M. It should be plug and play. :thumbsup:

Dell Alienware M18X M17x Nvidia Geforce GTX680M Laptop Video Card CPCXD 680M 2GB

 

@ J.Dre, unfortunately the 680m is NOT plug and play in an R3. 3D will not work and requires a modified drivers to install. The plus side is that on a 120hz model you don't have to mess with the bios to disable Optimus .

 

Even so, it's better than purchasing another 580M. Purchasing another 580M is borderline idiotic.

 

That's for sure, unless 3D is a big issue and the card is cheap enough to be worth trying? The 680m appears to be the 'gold' class GPU for the R3/R4

Just for note to OP: Do not look at the 780m for an R3, you also get sound problems that need fixing with that card...

 

EDIT
Removed potentially wrong information

 

Dunno if it can drive a 120hz display at boot though? It may have changed with the latest AMD cards but the earlier ones POST'd at 60hz and would fail the display (8 beeps) in a 3D model with no iGPU connected .

 

Good point, I forgot about that. I thought the fix was to get the unlocked BIOS and disable IGPU/set to PEG only?

 

Nah, that just bricks it - been there did that many moons ago!

That's the fix for a 60hz display. Since the 120hz models are already setup to only use the NV, only the driver mod is needed to allow the install to complete. No optimus and no bios stuff

 

Wow, thanks everyone for all the responses - decent diagnosis itself is a huge relief.

That it is a heating problem has been my theory from day 1, and I have done everything to try to keep things cool, including a desperate attempt at running the machine without the bottom cover, with an external cooling pad fan directly under the Graphics card heat sink (not intended to be a permanent solution, of course - just for testing). This seemed to mitigate the issue somewhat, but I have had bad episodes with this configuration as well. Yesterday I had the thermal paste replaced (the old paste did seem crumbly, not sticky, so I guess it was bad as well). Unfortunately this hasn't changed much. In an effort to reduce temps, I had turned Aero off and I also changed the refresh frequency to 60Hz - no love.

I did run with the 314.22 driver for a bit, (it was the oldest driver which happened to already be extracted to my C:\NVIDIA folder). That didn't fix the problem but I guess if it mitigates the issue (will have to observe for a longer time now), I will go back to it, less frequent failures are always welcome. What's the theory behind version 314.22 being better for this issue? (Sorry, I did try searching the forums but there were too many results!).

I also tried removing two of the four memory sticks that were easily accessible at the bottom of the system (I think the other two are under the keyboard?) that didn't help.

I haven't used the 3D in over a year, so I don't mind losing 3D at all. Is the patched driver for the 680M required only if I want 3D or is it required regardless?

One of the problems getting a new card is that I'm located in India - getting an electronic item shipped to India from the States (or any other country) is a huge PIA and returns would be next to impossible. Also I'm seeing the cheapest 680M's at around $420 (going up to $700) - is this the right price range? As far as getting another 580M, personally, I was very happy with the performance and wouldn't mind another 580M if it was significantly cheaper - most of the sellers for this are from China though (but I'm seeing a few items at ~$220). Besides ebay, is there no official source/store for these cards, other than asking Dell for a replacement (which I'm sure is going to be quite expensive)? Are there any other cards in between? Or are there cheap cards that I could buy just for testing and getting by until I can get a 680 from the states?

Thanks again for all the notes!

 

Yeah, sounds like constant overheating has finally killed it. 314.22 was the last driver before dell released a 320.xx. That release overvolted the 580m and that was all that was needed to kill a whole bunch if cards. Of course NV refused to even acknowledge the voltage boost and just played around until all 580 owners got fed up and moved on.

See this thread ( http://forum.notebookreview.com/ali...x-580m-78c-throttling-see-post-191-later.html) for the origins of the overheat. It's not just paste but the padding of the VRM's that causes a massive overheat. It's a strip of putty like stuff along one edge around 3-4mm thick. The fixes are moot anyway as time has shown it's a rubbish chip that requires perfect installation and cooling.

The modified driver is needed for the install to work but if you read anything about bios settings you can ignore that step on a 3d model. 120hz works fine just not the 3D. You can easily get the modded inf from laptopvideo2go.

India does seem to be a problem. I usually advise to simply buy the heatsink for an R4/680m as it comes with all the pads/paste in-place and only needs screwing to the card. Sourcing one is going to be tricky if dell never sold the r4 in India? Price looks areound correct. I got mine for £260.00 from a private seller - risky though...

There is nothing in-between. The 675 is exactly the same card as the 580m and getting that will just give you a similarly unreliable chip AND you'll need driver mods. A cheap 580M, IF you can be sure you have it padded correctly may well get you by for a while.

Think I've covered it?

Good luck.

 

Micky,

Thanks for the summary. I had heard, last year, about an Nvidia update bricking cards and I decided to stop updates for awhile. I resumed updates many months later, assuming they would have fixed the problem by then - I wish I had checked

So the failures that I'm seeing now are not because the card is overheating now, but because overvolting has already damaged the card and even though there isn't a heat situation now, the card is nearly dead?

It's going to take me quite some time to procure a replacement card, if possible at all. Need to look for friends returning from the US and then too, I have to hope I get something genuine - the 680M seems to be the best choice as you guys stated. In the meanwhile, is there anything I can do to mitigate the problem? Any way at all in which I could eek out a few months more out of the laptop? (Even older drivers, disconnect LCD, use only with external monitor, etc.)

At this point I don't even care about being able to games, all I want is a machine that stays on, reliably.

A couple of other notes:
1) I contacted dell asking for an out of warranty replacement. They said they would send me one for just around $80 (including the technician who would fit the card). This would be a 580M, of course. They state that this is a refurbished card and they will provide a warranty for only 90 days. Even so, this seems too good to be true - as long as I stick to 314.22, I guess the card should last for another year or so. The 90 day warranty should be enough to test the card (hopefully it hasn't already been screwed by the overvolting drivers).

2) I found my purchase information - I know I extended my warranty by two years but Dell claims it was just 1 additional year for hardware (so total of 2 years H/W) and a final year of Phone support. I bought the machine in Nov 2011, so I'm in the last few months of Phone support, with the hardware warranty having lapsed already. The machine was purchased in the States and after six months I had the warranty transferred to India.

I finally found my order details though and it's a bit hard to make sense of the warranty info (see below); however, doesn't "Dell Limited Hardware Warranty Plus In-Home Service After Remote Diagnosis, 2 Year Extended" mean I have a total of 3 years hardware warranty? Because it says "2 Year Extended" as opposed to "Extended Year" that you see in some of the other items. This would make a huge difference, obviously with free and immediate hardware replacements (right now option #1 above could take upto 3 weeks!)

Thanks for the great help so far!
Brian.


Warranty excerpt from my order summary
939-0330

Dell Hardware Limited Warranty, Extended Year

939-0320

Dell Hardware Limited Warranty, Initial Year

939-0380

Dell Limited Hardware Warranty Plus In-Home Service After Remote Diagnosis, 2 Year Extended

939-0340

Dell Limited Hardware Warranty Plus In-Home Service After Remote Diagnosis, Initial Year

950-3339

3 Year Limited Warranty

421-1721

Soft Contracts Service Hardware Agreement for Alienware

993-8582

Warranty Support, 2 Year Extended

994-3730

Warranty Support, Initial Year

 

My pleasure,

I'll answer in sequence of your post, hopefully it makes sense .

With the benefit of hindsight my take on the whole issue was the card was the last of the Fermi cores and like the 880m (last of the Kepler core) was overstressed at factory settings. That voltage boost which was clearly over the card's recommended max (even if a tiny amount) was enough to finish off a lot of cards. It should not have done, so now they were in a fix. NV only fixed it three driver releases ago, it just vanished, no acknowledgement whatsoever! So, the cards that survived have simply got to an age where they are failing naturally, if prematurely.

On mine, when I could not maintain windows for more that a minute or so I used safe mode. At least I could surf but that limits your use quite a bit.

You clearly (IMO) have hardware warranty. Scan that doc and ask to open a case with them, then sent it to the tech asking for a written explanation of how you are not covered.. When hardware is transferred between countries the receiving country often tries to get out of providing the warranty. It's a total loss situation for them, they never got the original cash - unless dell have some sort of internal cross-country recharge - unlikely.

For that price it's a no brainer though. 90 days is poor but they can get away with this in many countries and it will tide you over until you can get to the states. And as long as the 680m does not go into short supply it will only get cheaper.

As for the driver, you should be fine use one of the latest ones. Even 314.22 had the overvolt, it just did not stress the card and cause it to overheat/breakdown. It was after this version that Fermi core cards started failing.

Good luck with dell - they can't argue your receipt - guess they hoped you would not have it!

 

Micky, thanks again - so it looks like I should run with the latest drivers as it removes the overvolt too. Where do I actually see this in effect? NvidiaInspector reports voltages all the way from 0.8 to 0.92 - Do I need to observe what it is at a particular P state?

As far as warranty is concerned, I think this is very interesting:
1) First call to dell tech support: "No sir, I can assure you that '2 years extended' is the same as 2 years total. It's the difference between American and Indian English (!?!) If you have a dispute, you have to call Dell U.S.".

2) Second call to tech support (different guy, obviously): They first transferred my call to customer service for the warranty dispute. Customer service listened to me, but transferred me back to tech. Again, "Yes, it says '2 years extended' so that means 2 years total." I had to then point out that I had an identical item that said "initial year" instead of "2 years extended". He put me on hold for 15 minutes and then came back confirming that the machine did have three years hardware warranty! The good thing about this guy was that he was willing to listen and went to my US order page to check details there. I had a minor heart attack when he said that even three years had passed because the machine was purchased on 11th of August, 2011 (which I've now missed by 2 days). He was reading the date incorrectly ... 11/8 as 11th August instead of 8th November (India uses DDMMYY as opposed to MMDDYY).

So it took 70 minutes on the phone, but I finally won that battle.

The war, however, has not yet been won. He has placed a request for warranty extension against my tag, but another department needs to verify this - which could take upto 7 days. Until then, they can't process my service request. At this point, I hope they don't backtrack to the the previous "your machine has only 2 years of hardware warranty" stance.

I believe the problem started when the person transferring my warranty (way back in 2012) made the same misinterpretation, I guess this is a manual process. The lesson to take away from this is you'd better check dates and periods immediately after the transfer to ensure you don't get screwed when you need help the most. I started calling Dell about a month ago and was treated as an out-of-warranty customer. If I were in warranty, I'd probably have my card already replaced - right now I'm sitting here with a dead laptop and it's going to be at least another week until it's fixed.

I'm angry at myself for just accepting Dell's original 2 years of hardware + 1 year phone support claim without checking my records (would never think that their records could be incorrect, and I couldn't be 100% sure what we did when purchasing the machine 2.5 years ago). But I'm furious at dell for so nonchalantly dismissing my claims, which they would have gotten away with had I not argued with the "initial year" point. The magnitude of this is significant - first, I've paid more for the third year ... and now, I would pay even more for a new card (having already paid for a technician's visit which I shouldn't have had to). More importantly, I've already lost a month pottering around with this machine - makes my blood boil!

I'll keep you guys posted on developments - hopefully this time next week I'll have the laptop working again.

 

Hey, good for you sticking to your guns :thumbsup:.

Not there yet, but it sounds promising. I would chase with a daily call - no 'decision' should take longer than 24 hours. The problem is that support varies wildly in it's interpretation of the warranty terms & conditions depending on the local service. It's written and structured to allow local interpretation. Use of depot service comes to mind. If the remote diags identifies more than one part 'could' be the problem then they can ask for it to be sent in. In practice countries with larger volumes and higher expectations get an engineer with multiple parts. I am seeing a lot more US customers being asked to send their machines in so don't feel hard done by .

You make a good point for anyone that buys an alienware in another country. Simplest way is a support call. First thing they do is tell you your warranty status - decision tree branch one.

Laughed my a** off ' the difference between American and Indian English' These guys have all the answers. I was told that I couldn't transfer or buy accidental coverage in Mexico (from the US). Questioned this a little harder to be told 'It's cultural differences sir'. Ahh, so you're saying that there are countries that make too many claims? 'No, it's just what local people expect of a warranty'. Great PC answer.

If it wasn't for this interweb thingy most people would be happy with the level of service they get (but you are in the right here!).

It must be tempting to just fess up the $80.00 but since you already paid for a visit, this is out of order. There is an aw warranty resolution email but it can only deal with US and CA problems directly. They may be able to bounce something into local support to move things along? If I see it I'll post

Keep at it and good luck

Edit: forgot about the driver questions .

Use msiafterburner to display the clock speed and voltage (and any others that interest you like FPS) when gaming - settings, monitoring, in OCD. The stock max temp is 0.87v. 0.92v was tried by overclockers as a safe overvolt - big mistake and a high failure rate.
If you see 0.92v then it is overvolting. I'm getting one of those 'bad feelings' about the latest WHQL so I reccomend going one back - which seems to get a lot of decent press. I'll post the version when I've checked it out

 

The state now, is that my warranty extension actually went through without any further obstacles on Thursday last week. The support tech agreed to send a motherboard, graphics card and a blue ray drive (unrelated issue). Big win here Unfortunately, a couple of Indian holidays and the weekend have resulted in "next business day" = 5 days later. As you suggested, it was important to keep calling because they sure as hell weren't calling me. Even today I had to call to ensure that the tech visited tomorrow and not Wednesday, my parts were supposed to arrive this evening ... fingers crossed!

Actually the support period here made me sit up and research my warranty. They said that it could pretty much take one month to get me a replacement since I'm a lowly out-of-warranty customer, even if I paid. Unacceptable.

Would be sad if I get a new card and that dies as well I'm really interested in knowing what the "safe" driver to use is - and given this experience, probably will never update from it, ever

 

That is unacceptable. One month just because you extended a warranty for an out of warranty laptop?? Ridiculous. Once your warranty is paid for, you should be treated exactly the same as all other warranty holding customers!

 

Glad to hear at least they are going to fix it - finally!

The v314.22 has been used by a lot of people seeking a good working driver. The last but one WHQL, 337.88, is getting good press and a lot of users of the latest one (340.52) have rolled back to it, mostly 7xx/8xx issues.

Good luck, fingers crossed for the call-out

 

Lol do you not just love how dell tries to pull something over on you .

 

Current State
Graphics card replaced, system has been working flawlessly through nearly 20 hours of uptime right now. Before that I couldn't even get to the log-in screen. Given the intermittent nature of the problem, I'm refraining from jumping for joy until it works for at least a week, though I'm quite sure this has fixed it for now. The only way I see this going south is the new card slowly turning bad as well. I've reverted to the 314.22 driver but I haven't played any games yet.

Dell's Service
On the service front, Dell sent in a new GTX 580 but did not send any thermal paste! They did not send any thermal pads either, nor the thermal harness thing at the back. The engineer had to reuse the pads from my existing card, and it really looked like a mess. In fact, the harness thing at the back actually broke (the plastic bit) during removal and had to be reused in that form. He told me that the process at Dell is really screwed up and it's difficult for them to request additional parts. Luckily I had some paste left over from two weeks ago and he had to reuse some of the paste that was on the older card ... The resulting low quality of this work is my biggest worry wrt. longevity of new new card. Maybe I'll schedule a replacement (with the thermal assembly this time) just before my machine goes OOW in November .

To clarify, warranty extension (when I was considered an OOW customer) was impossible. This was the turn around time for OOW support. First I had to wait for them to confirm that the part was in stock, then make the payment and after that was confirmed, it could take 3 weeks.

In the end, I don't think Dell deliberately shortened by warranty - seems to have been a mistake during the transfer. However, the first few times I called them, I kept stating that I didn't think their information was correct (before finally accepting it), none of them suggested any remedial course. Finally, when I was actually contesting their claim, as you can see above, one of them just dismissed it out of hand - this is unacceptable. I also paid for the engineering visit, an inconsequential $20, but I'm irritated by the principle of the matter here. It was also a waste of time given that the engineer came with no parts and couldn't diagnose the problem at all (I posted here after the visit).

Micky, given that you said that 314.22 still had the overvolt but did not stress the card, I've gone back to that - but if there is a more recent driver that removes the overvolt entirely (you did say that the problem disappeared later but perhaps has reappeared now) - I would like to use that instead. It would be a more recent driver (with hopefully some benefits?) and without the overvolt. I know 340 has the overvolt and I think 337 did too. Other than trial and error, is there any chart somewhere where I can find the right driver?

 

Hi, some good news at last!

I am very surprised they did not send a new heatsink, especially as this card has already been identified by dell as having overheating issues. See the first post here: http://forum.notebookreview.com/ali...x-580m-78c-throttling-see-post-191-later.html. Dell supply the HS because of the problem of poor pad/paste usage by the tech. When I attempted to buy a 680m from Dell I was told that it only came with the heatsink and I would need to buy that as well to retain my warranty. Different countries though, and different processes .

See this post as well: http://forum.notebookreview.com/alienware/721878-nvidia-320-18-driver-warning.html for some version ideas.

I'm afraid it is some trial and error to find the best driver. I wish I could remember exactly at which driver it was that 'lost' the overvolt. I assumed it was gone for good once they identified why it was happening and the problems with the latest driver may just be something different. Start with 337.88 to see if the voltage kicks-up at game start. This is when it happens. It is such a short OV that it will not cause any instant damage, as you noted, 'over time' is the problem IMO.

You also need to game ASAP and monitor the temps as well as voltage. My average temp on that card when gaming (stock clocks) was around 68c and would rise to around 72c after a time. If it runs hot then it's back to dell for the heatsink. I wouldn't expect any argument since the tech reused the pads and the paste is unknown - all OEM's have specific requirements of thermal paste and although they are all much of a much (IMO) only dell supplied paste should have been used. But the service there IS a little wanting .

Test with Heaven 4 bench: Heaven Benchmark | Unigine: real-time 3D engine (game, simulation, visualization and VR)

That will warm it up and even dell techs use this one here in the UK (wherever they are!) to test overheating/shutdowns.

Thanks for the update and I hope it continues to work until you can upgrade later