M17x-R4 Black screen crashes ATI 7970m

While it happened only with Diablo 3 I am experiencing more and more black screen crashes while playing games.

With skyrim it is usually before 10 minutes, with DOTA 2 sometimes I can play a game others time it crashes at 30 min or so. The same with BF3.

I am starting to think this is a hardware issue but the strange thing is that I have been running Unigine heaven for 2 hours and it hasn't crashed so I don't know what to think.

Should I call the technical service and ask for another card or should I just wait for an official driver? I am using the ones that dell provides on his website.

 

Hello Alenghil, you could send an email to: [email protected]
Add your contact information and service tag so we can ask the guys from Gaming Support.
Also you could try to install the driver from AMD website, for that make sure you fully uninstall the previous driver; then, after the restart, install the new one.

 

I have the exact same issue with the AMD card. When playing games it crashes to a black screen, sometimes there's a buzzing noise, and I'm forced to shutdown holding the power button down.

I've ran 3dmark and it scores well. I ran Furmark and the gpu never got above 76 degrees. I ran Dell's diagnostics and it didn't find any issues. I've done a fresh install of the OS, grabbing the newest drivers available on Dell's site. I've tried the 7900 Mod driver.

I'm out of ideas and am thinking of sending it back to Dell.

 

Having the exact same problems here. Underclocked the cpu by one multiplier and it at least reduced the frequency. Would like to see a fix for this

 

I pretty much have the same problem.
So far im getting Black screen lockups in diablo 3 (e.g. belial last phase after ground AE) and guildwars 2 beta.
I've tried dell-drivers and a few modded ones from benchmark3d.com. memtest ran the whole night with no errors. I disabled turbo mode on CPU just for testing. - no change.
The only way i could fix the crashes, was switching to integrated graphics. The 7970m is definitely the cause of the problem.

 

It seems that if I downclock the card core to 800 mhz I don't crash anymore so it seems that I have a defective card.

 

alenghil, what are the temps of your card underload?

 

Is this running on the 7970 directly (Fn+F7)? Or using the switchable graphics?

If you are not running fully on the dedicated graphics, give it a try. The switching is not flawless yet.

 

The max temp I have seen is 77º with fumark or OCCT. It is normally at or below 70º while gaming.

I have been using the 7970 alone, I desactivated the HD4000 to test if it worked better.

PS: With the HD4000 I don't have any crash on any game.

 

I also crash a lot when overclocking on some games. It's usually certain sections that causes it. At stock, it's usually stable but I also had to downclock below the stock once or twice. It's like certain spikes are too much for the GPU to handle. I hope a driver update can fix this in the future.

 

Ditto to *all* of above. I've even ran the card on Furmark for 15 minutes straight at 975/1475 with *no* problems. 3DMark 11 runs flawlessly with 6750ish scores. At least once an hour on *stock* clocks it crashes in Diablo 3 in the same manner, though. Black screen, audio stuck buzzing. The only other game I've played so far is World of Warcraft, and extensively. No crashes at all. The frequency that it occurs doesn't increase as I increase the clocks, even at the 975/1475. I've noticed it happens most frequently when the elite skeletons that have fire trails on the ground appear.

I've also tried my own hybrid installation of the 12.4 drivers to no avail. Everything else still works great, in fact the clocks not staying at 2D clocks with screen-off power saving is gone. Have we considered that perhaps we're chasing the wrong thing, and maybe it's actually an audio problem? That would make more sense, seeing as the problem doesn't happen more frequently for me on overclocked settings.

EDIT:

I'm also running dedicated only. Integrated is completely disabled.

 

I've not had this issue on my M17x yet but I did have it on my old laptop (Gateway P7811FX). During games it would suddenly black screen with the audio looping and the only way to fix it was to hard restart. It would decrease in frequency when I underclocked the 9800m GTS. After a lot of research it turns out to be a dying graphics card... That's not to say all instances of this happening means a dying GPU, but I just wanted to share my experience.

 

Is this happening to every m17x or only some?

 

I recall that, too, and it was the same with the 8800GT because the 9 series were just rebranded. This seems to be more widespread and right out of the gates, so I'm hoping it's not a design flaw or whole batch of bad cards.

Testing a theory. I've noticed that this R4 is much more quiet than previous gens, i.e. the R4 waits until the last possible moment to ramp up the fans. I noticed this during benchmarking in 3DMark 11, where the fans wouldn't kick on full until close to the end of each test. I'm thinking that at certain points in some games like Diablo 3, there is a sudden burst of GPU usage in certain spots (like my freezing point right when elites with fire trails first come on the screen). Using HWiNFO64, I've set the fans to run at full speed 7K RPM at all times, and I'm going to see if the problem still happens. So far it hasn't, even with said elites coming out and even with the 975/1475 overclock (trying to force it to happen).

Will report back after testing through tomorrow or if it happens again and I try it with default clocks and it still does it.

 

Bumping up the fan speeds didn't fix it for me. I've already tried that a few days ago.
GPU temp is around 66°C in diablo (with auto fan control). So at least the core isnt overheating. With maxed fans it doesnt even get into 60s range.
Did anyone already repaste the GPU and checked if all thermal pads are in place?

The weird thing is that the card doesn't even run at full load when it crashes. D3 with vsync enabled requires about 60-70% as shown in msi afterburner.

Did someone try a modded vbios with at little bit more core voltage?
Black screens almost always happen if something is overheating or the card doesnt get enough power.

 

Yeah i now had this problem myself with 7970m. I found out that its overheating causing this. The GPU fan isnt active at all.

Im sure of this because i set the fans on HWiNFO and all blackscreen crashes are gone. Downside is that CPU fan runs at same speed as GPU = Unnecessary fast, which causes computer to make double the noise.

Any ideas guys?

EDIT: I am running 8,98 - 120604a drivers, and on my M18x

 

Same thing happens with M18x R1 with 7970M CrossFire. I'm inclined to think it is bad drivers. I uninstalled IDT audio and running native Windows 7 audio drivers the last 2 days. So far the buzzing and hard freeze has not resurfaced.

I have experienced the same thing with 6990M CrossFire and 580M SLI, although very infrequently with the latter, so I am looking elsewhere at the moment.

 

Whyyyy does this sound so familiar? I had an R1 and an R2 and so remember something like this happening before. I'll have to go digging through all my old posts.

I did post earlier in this thread that I thought it could be audio, though.

EDIT:

Driver problem was my other thought, too. That would suck because that will take time for AMD to push a fix out since even the current betas crash.

 

hmmmm, i don't have any of these issues with my R3. My 7970m has proven to be very stable (knock on wood). I'm running the modded 12.5 beta. I don't know if you guys can successfully install this one on the R4...it should install though if you're running dedicated. I can play D3 for hours without issues. It is usually a battlenet server crash that gives me problems

 

12.5s are "the only" non-Dell drivers that people can use. I've force-crammed 12.4s in because they ran so well with D3 on my Macbook Pro, but the same thing happens. Might try forcing the 12.6s next, but I don't think that's going to accomplish anything. I'm also considering the fact that it may be a voltage issue and might try 1.075mV overvolt.

 

U overclocking or stock? Your system will get stable at 1.075v, but i wouldn't go any higher. I'm guessing it is a voltage thing too

 

In D3, either way will black screen. Nothing else crashes, not even Furmark for an extended period of time on an aggressive overclock 975/1475. If it's voltage then it's a spiking issue, not a constant drain problem. I already went to 1.075mV a short bit ago. We'll see what happens. Extensive D3 session today. ^_^

 

You may want to try these....the latest 12.6.9 betas...and its already modded for 7970m

AMD Catalyst 12.6 9.00 BETA June 12 – already 7970m modded

 

I couldn't adjust the backlight brightness with the new drivers and therefore installed the old ones again.
800MHz core clock did the trick for me too. Looks like we got unlucky and our cards just need a little bit more vcore.
A official response from a dell/alienware tech guy would be appreciated if this issue can get fixed with a new driver/vbios from dell. Otherwise i'll try to get this card replaced when i can get a few days off from work.

 

I'm not having this issue with my R4/7970m. I did have 2 lock ups but not the black screen that you guys are talking about.

First lock up was in Tiger woods 08 running in compatability mode.

Second lock up was playing multiplayer game of space marine.

I have never had Skyrim, need for speed shift 2, or any of the other games that I regularly play lock up. I am using a cooling pad from cool master and have been watching the temps pretty close. My card does not seem to go over 68c even under load running stock clocks 850 and 1200.

 

Results so far with 1.075mV overvolt:

GPU overclock 975/1200 eventually black screened.
Mem overclock 850/1475 eventually black screened.
Mild overclock 900/1300 eventually black screened.
Stock clocks so far have not, but hasn't been a lot of testing.

All of those crashes took at least an hour of play to show up. Keep in mind - I was able to run Furmark 975/1475 for 15 mins straight with no issues, and 3DMark 11 multiple times on those clocks also without crashes. It's not CPU either - Prime95 stress tested that with no problems, either. I'm starting to lean toward driver problem. A lockup at stock clocks now will pretty much confirm either that or a wide-spread card flaw. Seems too odd that only specific games at specific points cause the crash.

EDIT - I'll probably give those 12.6s a go later tonight.

Played about 3 hours of D3 on stock clocks with 1.075mV overvolt, including running into the fire trail elites that would more often than not black screen me. No crashes. Just popped in the 12.6 betas that were released today, going to see how it goes. WoW for the next few hours, then back to D3. If all goes well with these, I might try the 900/1300 overclock again and see if the issue is resolved by the new drivers.

 

So I'm guessing this is only on some cards? And from what I read looks like a flaw in the cards but what do I know

Sent From My Rooted EVO 3D

 

How did you overvolt it? I can't get Afterburner to unlock the voltage bar, even after unlocking voltage control and monitoring in the settings. Thanks.

 

Don't take my word for it but I think you have to flash a special vbios of the voltage you want but slickdude would be the one to go to for this he has a thread on overclocking and overvolting the 7970m in the alienware topic section I think let me go find it for you
EDIT its in his sig lol I believe it doesn't have overvolting but you can ask him on sure he's would be willing to tell you how the others will to I'm sure
Sent From My Rooted EVO 3D

 

Just wanted to add for people without an AW looking to buy one that I have no issues of screen crashing. I have an R4 with a 7970. Only issue I've had with sound, it would disappear if I plugged in and out my headphones multiple times and sometimes the sound would sound muffled like the speakers were blown but a restart would fix it.

I do use a cooling pad though nothing expensive, just a $15 coolermaster.

 

I pulled, modded and flashed the BIOS with the overvolt. Just ran for 3 hours perfectly fine in WoW. Probably going to run 3DMark 11 now before I jump back into D3.

If this overvolt ends up working out, I'll post the BIOS for you to give a try.

Well, that was short lived. These 12.6s aren't overclock-friendly, at least not without forcing it in the BIOS. As soon as I set the clocks to anything over stock, I start getting a bad flicker every time the clocks change between 2D and 3D. The flicker won't stop until I set stock clocks again. Also, 3DMark 11 lost 100 points on the same clock rates as my modded 12.4s. I'll give it one more run but not looking good for anything beyond stock.

EDIT:

Same score the second time around. Between the performance loss and the flickering with overclocking in MSI, the new 12.6s aren't for me.

EDIT 2:

After closing and opening MSI, the flickering isn't happening now. Overclock still in place. Odd. Will dig a little bit more into it.

 

I installed the 9.00 catalyst beta ones and the fans seems to be kicking in properly. BUT Deus Ex Human revolution wont run maxed on 1080p, neither wont Crysis 2. I get bad performance on those, specialy with v-sync on. Either the 7970m aint that powerfull or the drivers really suck.

 

Uninstalled the 12.6s and "clean" installed 12.4s using the 12.5 installer. My previous 12.4 attempt was simply overwriting the driver binary files, but it had a few residual 12.5 files still being installed, and 12.5 config settings. With much more clean 12.4s installed, I was able to run Futuremark again without problems. About 45 minutes into a D3 session it went to the black screen again, though. This pretty much eliminates the drivers as the issue, unless it's a problem back to 12.4s as well. Is anyone on the stock Dell drivers getting the black screens?

 

The first black screens i got, occurred with stock drivers... thats why i tried the beta drivers.
Did D3 crash with standard core voltage or 1.075V? What clock speeds did you use when you got the black screen again?

 

Ok so it looks like I might be sticking with the newest 12.6 betas after all. The performance boost in Diablo 3 even on stock clocks is enough for me to keep them. With previous drivers I had to run RadeonPro to keep the framerates consistent without stuttering. I guess they worked that problem out in the new betas and it's no longer an issue. I'll need to do some more testing as far as overclocking goes, and whether that flickering occurs again when I do, but I was able to run Diablo 3 for the past 3 hours with no crashes at all. I'm still overvolted to 1.075mV, but perhaps this is the magic combination. After I test tomorrow some more I'll go ahead and post the BIOS if people want it (do at your own risk, of course).

 

I have M18x R1 and am having the same issues. On 12.5 beta I had less issues and plan to go back. I have had the black screen in The Witcher 2 and Diablo 3. A mild overclock 900-1250 isn't even stable without that issue. I get it on stock clocks which I am on now. I did try the overvolt and had more issues than on stock volts. Now I have tested without even the overclock options enabled ie "extreme edition" disabled and bios at stock. I still got black screen. It's not cpu/mem related. I play diablo 3 no vsync so it is rendering more than needed frames but should not cause issues. This has been a good test bed and Diablo hates 7970m. I have a crossfire set up but entire use of it's been in single gpu mode. So Xfire isn't the cause and obviously not for you R4 owners either. Now I normally game with fans maxed to pre-emtively keep all my components cool so heat doesn't have chance to build up. For my testing the crashes are occuring now more frequently when I hit 74-76c range basically where the fans finally kick in to high... But prob is they hit nearly 75-76c before even going to medium fans and this is on bios A05. When it crashes it's been after a bit of heat builds. With fans on max I am seeing less issue if any. I will continue to test at stock clocks and especially stock vbios. Perhaps the big probs are from the die shrinking so much. The same heat last gen could handle may be way too much for these 28nm chips?!

On overclocking testing I had no issues running benchmarks but if mem or core got to high the black screens occur to. I think those who r having black screens and overclock need to reset to stock clocks and retest to see what's happening. On a light overclock I could crash at 62c with fans on max :/ so again I'll be playing more Witcher 2 and Diablo 3 on stock clocks with no fan control. When I gane normal I'll go back to my preffered max fans but to test our cards on stock clocks for stability... using fans on max shouldn't be a necessary step. The EC should probably have more pre-emptive values, but to determine if this is a widespread issue stock clocks gpu and stock fan control should be used. I feel these would be less frequent if high gpu fan kicked in at 65c vs 75c... a temp that is technically worlds apart. My previous gt580m were unstable if they ever hit 78c under any heavy game such as Witcher 2 so I don't see these to be different. Not that each 7970m will crash at mid to upper 70's under such games I'm sure people have had them into 80c... but seems directly related to gpu speeds, voltage and temps. As with most processing cores if heat is raised the required voltage core to keep stable may be higher during use. Perhaps at as temps start to raise over 65+ the gpus may need slightly more volts like some of you suggested and since all silicon is different maybe some need more volts at stock clocks to stabalize. In attempts though I did not see anything I would call gaming stable on any of the overvolt bioses.

So stock gpu and vbios here and today I'll go back to 12.5 beta and continue testing.

Btw on 12.6 I had the flicker on gpu clock change to in afterburner. I hit reset no flicker add a single mhz flicker... All I did was flash back to stock vbios and then afterburner hasn't done that again...

 

Can someone without problems on the 7970m do a test with memtestcl?, I am getting errors on random blocks in 50 iterations testing only 128 MiB of the GPU ram, the rest of the test is fine. Maybe the software is not updated to work with the 7900 series or maybe thats the issue of our cards.

By the way, I haven't had any crash at 825/1200 with stock voltage and 12.5

 

where do i get this memopencl?

 

Ups sorry it is called memtestcl I don't know what I was thinking when I wrote it.

http://fah-web.stanford.edu/MemtestCL/memtestCL-1.00-windows.zip

 

Tested,

Final error count: 0 errors

 

I had the black screen crashes and I don't get any errors running that. Then again, I just ran Diablo 3 for over 4 hours without a crash. My magic combination was new 12.6 betas, the 1.075mV overvolt, and stock clocks. Not sure which of the first two, or both, has made me stable.

 

I had blackscreens but cannot remember with what drivers. Might be 12.4 or 12.5.

Now im running 12.6 and its all fine.

Could you all who have blackscreen issues to closely watch the behavior of your GPU fan since i noticed that it was running in idle speeds when temps of GPU was over 70´s. I think that is what caused the crashes, overheat.

 

So if you guys don't get any error on the test and I have 26 millions of random block errors on 50 iterations I think the mem of my GPU is bad, I will call the technical service and ask for a replace of the GPU as soon as I am back in my house.

Has anyone else with black screen crashes got any errors on the test or it's only me?

 

I forced the fans at max speed yesterday - with a GPU-temp around 56-59°C in d3 and still had a crash at stock clocks. I don't think heat is an issue at least on the m17xr4.
I'll try memtestcl when im back home.

 

My fan is running at 2500 rpm at 57º and at 66º is at 4200-4500 rpm, so I don't think I have a fan problem.

 

Oh, cool. Then i can totaly rule that one out . Thanks guys.

My computer is weird. The GPU is idling at 50-56C with 0 (Zero) fan rpm.

1800rpm at 65C
2000rpm at 70C
2300rpm 72C
2800-2900rpm at 74C
3400rpm at 77C
3600rpm at 79-80C

Only got it to 82C

Now i switched to x-treme burn and its running 86C with 3600rpm , it wont go higher.

THOUGH i did make another discovery, i see my GPU clock jumping up and down from 450MHz to 850MHz, do i have throttling?

This is a very quiet machine, i like it

 

Hello there guys, my first post here but long time reader

I am very interested by this thread having myself purchased a new m17r4 equipped with 7970m video card and seeing some weird problems while in-game and also during normal use...

The computer just randomly stutters while playing games be it Battlefield 3 (i've done some research about stuttering for this game...), Skyrim or Rift which are the only games i'm playing at the moment. But the funny (not so much...) thing is i get the same wierd stutters while browsing and listening to music or watching a video on youtube for instance...
If listening to music on youtube or deezer for instance and browsing on an other window then the computer would sometime just slow down so much that even the music is slowed down (as though somebody would alter speed on a vynil record by pressing their finger on it if you know what i 'm saying).

I am trying re installing audio drivers and updating to beta AMD drivers also this afternoon, and will let you know. Other than that if any one of you has any suggestions please let me know.

Thank you much!

 

Standard settings of memtestcl didn't show an error, but: "memtestcl.exe 1024 50" showed lots of errors in the same section you got them (random blocks).
Underclocking the core didn't help and i wasn't able to set the memory clocks below 1200MHz...
I tested stock dell-drivers, 12.5 and 12.6 betas all of them showed the same problem.

 

if it is jumping up and down during test then it is throttling, also 86C is quite a lot my 7970m never exceeded 75C

 

It never exceeded 75C in Extreme burn mode? How long did you run it?