** Video Memory Corruption ** M170 and now M1710!

Hi,

Towards the end of 2007 I was having video corruption with my M170 (7800 GTX card). There were lots of artifacts and lines even at the POST Dell boot logo. I ran dell diagnostics and it identified a problem with the video memory (wrote and expected 'x' from memory but got 'y').

I called Dell and they replaced the motherboard, LCD, graphics card three times but the corruption eventually came back every time.

Dell recently sent me an M1710 with 7950 GTX card. After 3 weeks of use the video corruption has come back! This time I can't even boot into windows (NMI Parity Check Error).

The system is under warranty but I sense that I may be going around in circles.

I've seen the other threads related to this issue but replacement graphic cards seem to work -for a while- and meet the same demise.

Any help or suggestions will be greatly appreciated.

Some other links for people with the same problem:
http://www.dellcommunity.com/suppor...?board.id=Tech_Talk_XPS_Laptop&thread.id=3011
http://www.dellcommunity.com/supportforums/board/message?board.id=insp_video&thread.id=171549p
http://www.dellcommunity.com/supportforums/board/message?board.id=insp_video&thread.id=171547
http://www.dellcommunity.com/supportforums/board/message?board.id=insp_video&thread.id=171499
http://www.notebookforums.com/thread209845.html
http://www.notebookforums.com/thread203491.html
http://forum.notebookreview.com/showthread.php?t=51748
http://forum.notebookreview.com/showthread.php?t=209408
http://www.notebookforums.com/thread210944.html
http://www.notebookforums.com/thread210914.html
http://www.notebookforums.com/thread210535.html
http://www.notebookforums.com/thread206665.html

 

Are you by chance running the laptop with the lid closed on a bed or carpet a lot? Video cards really only die that often when they are being stressed and overheating.

 

No, every time failure has occurred the machine was placed on a desk. I also run i8kfangui to make sure it doesn't overheat.

The latest video card failure occurred when 30 seconds into COD4 so I'm not convinced that it is a heat issue. But I may be wrong.

Help please!

 

Also, do you overclock? Stop doing it if you are.

NMI Parity Error could mean something is wrong with the RAM as well. Not sure though. MemTest86+ is much better than Dell's utility.

No offense, but are you reinstalling the parts (especially the GPU)? Is the person repairing it using a proper thermal pad or AS5? Could the GPU be coming slightly loose inside the system.

If there is no proper heatsink, a modern IC wouldn't need much time at all to fail believe it or not.

I remember a huge thread in the Dell portion of the forums about the NMI Parity Error that was seen on many laptops for a time. There might be a solution found in there...I'm gonna try and find that thread right now.

EDIT: http://forum.notebookreview.com/showthread.php?t=92036

 

I have never tried overclocking with any of my video cards. I have never really needed to do that.

I ran Dell diagnostics and Windows Memory Diagnostic Tool (from the OS selection screen) and the RAM passed.

I've never needed to open the Laptop since it is under warranty. All previous GPU replacements were done by Dell service engineers. Everytime the card worked for a while (about 1 hour into a game) and then the screen corruption appears. I doubt that the GPU is coming loose everytime. Besides I'm getting errors with video memory check which indicates something is wrong with the GPU memory. However, I can't identify the root cause.

I've looked through

http://forum.notebookreview.com/showthread.php?t=92036

I can't find an answer since I'm getting video corruption from the POST stage in addition to the NMI parity bit error. Others that have posted in the thread seem to get NMI parity error now and again and have resolved it by changing the drivers for the WIFI.

I think my case is different.

 

Hmmm...not sure then. I was thinking maybe Dell sent a tech to your location and they screw up (I do not have much faith in at-home service anymore).

Still try MemTest86+ though...it is a much more capable testing utility for RAM.

At this point it sounds like you are having way too many problems and you should not be. You might want to ask Dell to replace what you have now with the current XPS 17" machine they offer. Especially since it seems that they cannot fix it.

 

I didn't really want to go down the path of replacing my replacement machine.

I've also seen in other threads that people are having similar problems with Dell replacing card after card.

I was hoping that someone could help me identify the causal factor(s).

I'll try running MemTest86+ and will post back with the results.

 

Is there anyway for you to check the outlet you are using?

 

I'm not sure how to go about checking the outlet. I should say that the with the M170 the video card went when I was at home (mainly). This latest episode occurred when I was in the office.

Could it be the power adapter?

 

I *think* I have found the root cause of the problem.

In many threads, video\screen corruption is often seen when the graphic card overheats too much. This can occur more frequently if you overclock your card.

What I have noticed today is the following:

I have left the computer idle for about 5 mins. During these five minutes the temperature gradually rose to about 71 degrees without fans turning on.

At 72 degrees the fans turned on at 'slow' speed.

Last night the chain of events leading upto video corruption is as follows:

The M1710 was left idle for about 5 minutes.

WITHOUT setting "fans always on" in 18kfangui I loaded up Call of Duty 4.
Within 20-30secs the screen became corrupted and the system reset.

The GPU video ram is now permanantly damaged.

My hypothesis is that the fans (which are controlled by the bios) were unable to reduce the temperature fast enough.

Only problem with this theory is that the temperature must have been over 100 degrees to cook the video RAM.

Is there a BIOS bug preventing proper cooling of the system?

The BIOS version in my machine is A07.

 

Have you cleaned the system since you got the system?

It's pretty easy to open it up, and clean the dust bunnies out of it.

It just may be causing it, and wouldn't hurt to get it done.

 

No need, it is only 3 weeks old!

Good idea though.

 

In three of the four previous cases of of GPU RAM burn out, I noted that screen corrpution (VRAM damage) occurred when running a graphic intensive game (COD4 and ETQW). It is in these cases when the temperature needs to be controlled properly to prevent the video RAM becoming damaged.

I am starting to believe that it is a cooling issue specifically related to the BIOS being unable to appropriately cool the system.

I am starting to see too many people in this forum (and the Dell forum) with the same problem.

Dell's answer: keep replacing parts and then issue a replacement.

Dell needs to thoroughly test the BIOS cooling software and temperature sensors because it seems to be an epidemic amoung certain users (gamers).

Comments?

 

Was it new though?

If it was a refurb.. you should look into cleaning it.

I've seen system with idle temp of 7XC going down to good as new 5XC idle speed after cleaning.

Oh and I use i8kfan to monitor, not control fans. I let the bois decide on that one.

 

There doesn't seem to be dust in the system.

I think it is a new system but I'm not sure. How can I check if it's a refurbished machine.

I'll hoover up the dust in the system anyway (I don't want may new card to expire just yet

EDIT: I'm pretty sure that it is a new system. I remember the support agent manager picking and choosing every component within the system. Furthermore, it had all the transparent stickers and lid protection.

If it was a refurbished machine then it must have been sent back within the 7 days period that people have to check the product.

I've tried removing the dust but nothing much has come out.

 

I doubt that a system can idle and 50C odd without ever kicking in fans. The components constantly emit heat so the temperature will rise.

In my case it took five mins for it rise from 45C to 71C.

Note that the fans only kicked in at 72C (at slow speed)!

Again, I have my doubts about the BIOS control over the fan speeds.

 

Thing is though that video RAM isn't typically cooled by the heatsink/fan in a laptop - I know that's the case with the Latitude D series and Precision M2300 and M4300 - maybe in your case the GPU RAM is attached to the GPU heatsink since it is a higher performance part.

 

OK, so if the GPU heatsink is attached to the GPU RAM then how does this help diagnose the problem?

UPDATE: Dell engineer came around and replaced the 7950GTX (rev. A00) in my M1710 with an A02 revision. I asked if he has seen similar problems and he said yes. Someone had his card replaced three times until he got an A02 revision board. Fingers crossed.

It now seems that it *may* be a faulty batch of GPU cards!

UPDATE 2: I have been putting the new card through its paces (UT3) and it is performing perfectly in Vista. I fired up BF2142 in XP and I started to get screen flickering toward the end of the "optimising shaders" screen and I quickly turned the computer off at that point. It looked quite worrying and I hope hardware failure isn't around the corner.

UPDATE 3: I've been testing some more in Vista and all seems well. I now have to be dared to load up BF2142 in XP again!