M17x R2 GPU issues...

[Guest bifnewman]

@hopelesslyfaithful

My M17x has been having some issues lately. First my CPU was overheating, so I was able to repaste that and get it up and running better. Then, my 5870's were overheating, severely.

After repasting the GPU and replacing the thermal pads on the vram with thermal paste, the temps are much better. I hit around 82c on full load.

However, now I am seeing some artifacts in games like Skyrim and BF3, where the entire screen looks like its scrolling down while its drawing new frames. I just had a chat with a Dell rep; and btw, does anyone have a number for rep that isnt completely stupid? The lady told me that the column to the left are the values, the center are the mins and the right is the max. She said that 133c is okay. I asked her if she realized that that is 33c above the BOILING POINT of water. she said that we only care about "the values." She must not realize that the left column is the current temperature, not the average temperature. Anyone else think that the card is permanently damaged? I didn't take a picture, but I also had GPU-Z and HWinfo64 giving a reading of 120c, so I really don't think that the readings were wrong. I'm not sure how that would be possible, but then again, I am not sure how the system was still on if the gpu ran at 133c.

Any recommendations?

http://youtu.be/x_ImuCUc9-A

[StamatisX]

Normally it should have shut down, don't know why it didn't in case these temps are correct. In any case I agree those temps are insane and most likely damaged the cards so I would just insist on GPU replacements.

BTW did you run any other benchmarks to check temps and for artifacts like 3DMark11 or Unigine ?

[Guest bifnewman]

Normally it should have shut down, don't know why it didn't in case these temps are correct. In any case I agree those temps are insane and most likely damaged the cards so I would just insist on GPU replacements.

BTW did you run any other benchmarks to check temps and for artifacts like 3DMark11 or Unigine ?

Just ran Unigine Sanctuary benchmark, and it had artifacts. Seemed as if the screen was wiggling back and forth while it was running. Max temps are like 75c though. Had it on the highest settings, dx11, full aa, and everything.

[StamatisX]

If the drivers are the same like before then we can safely assume that the cards are fried...

[unreal25]

I'm not sure what "scrolling down while drawing new frames" looks like (I'm guessing it doesn't go away when you enable v-sync), but I imagine these games worked fine previously? What you can do is try running them on each card separately first and then if those issues only occur with one, that one might be damaged. If your computer is still under warranty you should then ask them to replace that card.

Otherwise, you could try reinstalling drivers -- I've had so many issues resolved for me by reinstalling AMD drivers, it's not even funny anymore.

[Brian]

At those temps the card is likely fried and needs replacement.

[Guest bifnewman]

Does anyone have a number for a Dell rep that will actually acknowledge a problem with this system? I really don't want to run into another person who is going to tell me that it is okay for my graphics card to run at temperatures above the boiling point of water. I was going to tell the lady that I repasted the CPU and GPU's, but I was worried that I would void my warranty. When we ran benchmarks and stuff, she said my computer was fine, which was true, except for the artifacts.

UPDATE: @unreal25 - trying a reinstallation of drivers right now.

UPDATE 2: Just played bf3 for around 45 minutes. Performance seemed like it degraded over time. At the beginning, it was running 60-50fps. By the end, it would dip into the 20s and 10s. Artifacting continued. The artifacting is hard to notice, but I still think it is a problem. Vsync is on btw.

[unreal25]

Yeah seems like a damaged GPU, like Brian said. I sent you a PM, I hope it helps!

[Guest bifnewman]

Uploaded a quick video of the artifacting!

Will add it to the original post.

[Brian]

That could be a defective Crossfire cable causing this rather than a dead card. If you got a spare cable, try it or better yet, disconnect the CF cable, and test each card separately in the primary GPU slot to see if the problem goes away.

[Guest bifnewman]

turning off crossfire makes this issue go away. i do not have a spare cable. it is possible that it could be partially unplugged. don't i have to reflash the secondary card to the master vbios to run it as the master card?

EDIT: i believe i have fixed the issue. the crossfire cable had a bent pin on the connector. it took me 2 tries to bend it back. thanks @Brian. far cry 3 runs like shit in crossfire, though. possibly due to drivers? i'll try to install a cap.

EDIT 2: i'm actually not even sure i fixed it. far cry 3 just runs like shit in general and so does skyrim. i wish this would just work...

[Brian]

@bifnewman if its not scrolling anymore and crossfire is active (check the CCC) then its likely drivers. The 5870M Xfire is getting a bit outdated for these types of games, might want to consider selling the 5870s and grabbing a single 7970M.

[Guest bifnewman]

@bifnewman if its not scrolling anymore and crossfire is active (check the CCC) then its likely drivers. The 5870M Xfire is getting a bit outdated for these types of games, might want to consider selling the 5870s and grabbing a single 7970M.

I was hoping to upgrade laptops before then... What would my notebook even sell for?

[Brian]

I was hoping to upgrade laptops before then... What would my notebook even sell for?

Hmm I'm not sure, if you have a 920/940xm then maybe $800. Unfortunately these systems depreciate very fast.

Sent from my GT-N7000

[unreal25]

Look it up on eBay, based on what people are bidding I think it could go for more than $800ish.

If I remember correctly, 7970M is about $500. Not sure for how much 2x 5870m sell, but maybe you could get $100 back there.

I'm not even sure what would be a worthwhile upgrade except for the pricey M18x with 680M SLI.

[Guest bifnewman]

you'd think i'd be able to run skyrim with no lag spikes?

[unreal25]

I don't know... I was never able to get Skyrim to run properly on 5870Ms. The game seemed to run fairly smoothly (I think it was some combination of medium/high settings on 1920x1200) but I couldn't figure out how to get rid of the artifacts I got with crossfire enabled (it worked fine on a single card but I had to lower the details). So, I eventually gave up trying.

[radji]

I'm confused. The problems all go away when you turn off crossfire, yes? That could point to two issues: one, your crossfire cable is faulty...which you said it was. You can get a replacement crossfire cable for pretty cheap off of ebay. If that doesn't fix SOME of the lagging (won't fix all especially in newer games), then the secondary 5870m card may be having issues. The only way to check that is to take the secondary card and install it on the primary GPU's heatsink and in the primary MXM slot. If you still get high temps, artifacts, and/or lagging, you then know it's the secondary 5870m. Before you go swapping out the GPUs, take out the secondary card and leave the primary 5870m card and see what temps that card gives under load. If the primary card seems to run fine on it's own, it's probably not been damaged much by the extreme temps. If you do go with updating the crossfire drivers, I would recommend you do a clean install. Uninstall all the AMD software from your system. Then restart and go to the device manager. Select your AMD GPU from the Display Adapters, right-click and select uninstall. Make sure to check the box to delete the software driver files after they uninstall. Then restart again. You will need to repeat the process until the AMD GPU shows up a Generic VGA adapter under device manager. NOW all the AMD drivers are gone. You will need to initially install the stock Dell driver package for the 5870m crossfire cards to work correctly again. After installing the stock Dell 5870m driver package, make sure both your 5870m cards are working in crossfire like they should. If they do, you can update to the latest AMD Catalyst drivers.

[elkay]

Ouch on the temps. You likely fried a card and will need to get a complete replacement. "Technically" you could run cards to 100 deg before you start having meltdown concerns, but anything over 85 deg is generally considered bad. I've fried NVidia cards only being in the upper 80's, but those were the old G92 cores that were notorious for overheating and failing.