Alienware m17x r3 nvidia driver BSOD

Hello, I have recently installed Windows 10 on my Alienware laptop. Everything was running smoothly until I installed nvidia graphics drivers for my GTX 460m, at which point the BSOD's began. I have experienced these BSOD's on Windows 7 and 8/8.1 as well.

What I've done in an attempt to fix the problem:
-Install different driver versions (355.60 and 306.97; clean install each time)
-Disable nvidia streamer service
-Update BIOS/vBIOS
-Format Hard Drive
-Check RAM (via memtest; 0 errors)
-Run sfc/scannow (errors in CBS.log)

The only way I have been able to prevent daily BSOD's is by removing the nvidia drivers altogether, eliminating any chances of playing graphic intensive games (also, I would prefer not to use an older driver because some games require the most up to date). I'd like to believe the problem is software related because I can boot into safe mode without issues. I have, however, been considering purchasing a new graphics card (probably the 780m upgrade kit) but would be moot if it doesn't fix anything.

Below I've attached several BSOD files, my system information (.nfo file), CBS.log file, as well as a couple GPU-Z Sensor Logs (for each nvidia driver version mentioned above). In the sensor logs, the BSOD occurs right after all values are set to zero. This is something I've been fighting for years, so any and all help is greatly appreciated.

https://onedrive.live.com/redir?resid=6E6984AA2D967F21!2743&authkey=!AMQLLVSn2D5G1cY&ithint=folder,dmp

Thanks!

 

Err, what GPU do you have? A 580m by any chance?

 

My graphics card is the GTX 460m (I've made appropriate changes in the original post).

 

Thanks. I think it's safe to say that the GPU has passed on. Had it been the 580m then there would no point in even trying anything else. That card was so unreliable I've seen at least 10 fail in the same way.

The symptoms are typical of a GPU failure and the fact it works fine on the on-board GPU says that the machine is otherwise OK.

There is one thing left to try, and it's going to be a learning exercise so even if it does not help you will be in a better place to carry out a GPU replacement .

You want to try cleaning the edge connector of the card and see if that helps. Remove the battery and underside panel. Then the GPU fan. After that remove the two card retaining screws and lift the card to around 15 degrees from horizontal. If you have ever installed memory then you know how these slots work. Once it is lifted you can pull it from the connector. It does stick and can require a little wiggling to remove the complete unit, card and heatsink. Do not undo the four screws holding the card to the heatsink. This is not needed and just makes the job more complicated.

Clean the edge connector with alcohol and blow out the slot with a can of compressed air. Re-assemble and test. If it's still the same then time to invest - unless you want to try a 're-flow'? For the desperate, heating the card to re-flow the solder has been claimed to fix an old card but since you are due a decent boost to your gaming performance I think this is not worth the effort?

Any replacement is going to require you to mess with the drivers and you will need to track down an unlocked BIOS so you can make the necessary changes to switch off the on-board GPU. Otherwise it will not work correctly.

Personally I would recommend the 680m in an R3. It's going to be a huge increase in gaming performance and runs nice and cool so can be pushed a lot - easily up to 780m performance. The 780m also sometimes messes up the on-board sound on the R3 and this can be tricky to fix so playing safe see if you can find a 680m.

You will find a bunch of threads on here detailing the work and parts required so hunt them down and then ask me if you have any questions .

Oh yeah, if you have a three pipe heatsink along with a 240w PSU then you can save a few beer tokens...

Good luck.

 

Thanks for the quick reply! This is actually good news, as I've been looking for a reason to upgrade. I did as you suggested, made no difference though. I intend to purchase the new card from eurocom ( http://www.eurocom.com/ec/upgrade(1,262,0)ec). Fortunately, my laptop came with the higher wattage power supply as well as the three pipe heatsink, so I won't need to buy those. I also have the unlocked BIOS installed already.

Couple of questions though:
1. If I go with the 680m, will the 2GB difference in memory make a huge impact?
2. Will it be compatible with Windows 10? (This will help me make my own .inf files: http://forums.guru3d.com/showthread.php?t=377158)

Also, I found a discussion on techinferno that details a fix for the on-board sound:
http://forum.techinferno.com/alienware-m17x-aw-17/4132-alienware-m17x-r3-upgrading-gtx-780m.html (from j95, step 1)

Thanks for the help thus far!

 

You're welcome .

For 1, When I researched it the general opinion was it makes little difference if you're not intending to use multi-monitor. I'm not sure that is true now for the latest titles that load huge texture files during play?
2. You will need to disable driver verification as even a modded inf is detected as an unauthorized package in win 10 but apart from that I would not expect any extra problems over and above the terrible NV drivers currently available. I have always got my inf's from laptopvideo2go/drivers. Ignore the search, no idea why it is even there, just select the driver version you want to use and follow the link for the inf. This can be applied to the original NV install.

I've got a bunch of OV Vbios I used on my 680m and found it runs so cool I can run the max overvolt and +220 mhz (stable) on the core easily. Temp never exceeds 73c!

I've seen a lot of j95's stuff and he's a sharp guy. Also helpful if you get stuck if you go for the 780m and have any problems.

Good news re the HS and PSU, and I do like the packages Eurocom put together. They may be a little more expensive but I found their support to be on-the-ball.

 

Alright, makes sense. Impressive overclocking! Would you mind providing a link to where you got your Vbios'? It sounds like I may not need anything more than the 680m. I'll let you know how installation goes, hopefully within a week's time.

 

I'll PM you a link to a package I put together for 680m users. All work credited and a donation link if you use any of the vbios files...

 

Got the card installed today. Everything was working great until vbios flashing, which resulted in a black screen on startup. I attempted to revert back to the original bios (using integrated graphics), restarted, and got 8 short beeps (black screen again, but this time no motherboard bios either). Apparently this is an LCD error, but it goes away when I remove the card. Advice?

Edit: I think I found the problem...my backup vbios didn't save properly. Is there any way I can recover it or find a version online? Here is the warning that may have led to this whole mess:
Firmware image PCI Subsystem ID (1028.0551) does not match adapter PCI Subsystem ID (1028.04BA)

Here's more info:


Should I try to flash from USB on boot?

Edit 2: Turns out I had the wrong vbios all along. This is what I should be using: 80.04.67.00.01
Everything is back to normal now! The strange thing is, this is a 2GB card, but the vbios is for 4GB...So now I have a new question: Where can I find modded vbios' for this version? (Clevo P370EM)

 

Is that from the files I PM'd you a link to? I've used all of them on my 2gb version. From what I read some time ago the 680m vbios works on either memory model, not the case for some other cards though.

The mismatch is normal. maybe just a bad flash? Just make sure you make a backup of the working rom before trying others and just start with the stock voltage one and see what OC you can get on stock voltage...

Edit: Forgot to add, the 2gb version was only made for Dell but works fine in other machines.

 

Tried reflashing to links you gave me, still didn't work. Reverted to original and now my screen shows a strange color scheme that flashes frequently. What does this mean?

Edit: Fixed the weird colors by discharging everything (removed cmos battery). Still haven't found a good flash for tweaking voltage.
By the way, I found original vbios here: http://www.techpowerup.com/vgabios/133470/clevo-gtx680m-4096-120912.html
Couldn't find a modded voltage vbios on that website though.

I'll try modding without over-voltage later today

 

That was strange - draining the cmos should have no effect on the GPU, maybe you reseated the card at the same time? Dunno.

I'm wondering if it is a 4gb model and does needs a different vbios? I'm fairly sure I got them from techinferno, maybe a hunt there?

 

That makes sense; I did remove/replace the card and then drain the power.

Found a link on techinferno: http://forum.techinferno.com/genera...-editions-modified-clocks-voltage-tweaks.html

svl7 posted a variety of vbios'. I'm thinking of trying Clevo 680m - 80.04.33.00.10_'OCedition'_revised_01 - OV 1050v.zip. Problem is, I don't have access to download from techinferno (I've never posted there). Could you download all the Clevo versions for me?

 

Thanks! It flashed flawlessly! Overclocked GPU +220 MHz @ 1.05v, ran heaven benchmark on extreme settings and got max temp of 80 C. Also ran Kombustor and got a max temp of 90 C. Is there a need to overclock memory?

 

I've never actually tested the benefit of memory OC. I usually bump it up around +150~200mhz just because I can. When probs occur memory is the first thing I slow down.

Temp is a little high but you are running the max voltage and a serious overclock. I would imagine it will be considerably less in actual gaming.

Good news!

 

Sounds good. I'll monitor the temp for several games just to be sure. Seems everything is fixed...for now. Major thanks for the help! I'll be back if I have any more issues.

 

Cool, I just noticed you used Kombustor. Don't, if anything is going to push a card too far that will!

 

Noted