• Some users have recently had their accounts hijacked. It seems that the now defunct EVGA forums might have compromised your password there and seems many are using the same PW here. We would suggest you UPDATE YOUR PASSWORD and TURN ON 2FA for your account here to further secure it. None of the compromised accounts had 2FA turned on.
    Once you have enabled 2FA, your account will be updated soon to show a badge, letting other members know that you use 2FA to protect your account. This should be beneficial for everyone that uses FSFT.

Continued GSOD problems with AMD

S

suibhne

n00b
Joined
Dec 28, 2009
Messages
8
I've seen folks say these issues have to be related to hardware problems on the customer's end, now that ATI/AMD has fixed all of the other problems via drivers, so I invite you to tell me where my problems lie. I'm at my wit's end.

I just got a replacement 5870 from XFX via RMA, after wrestling with troubleshooting my old 5870 for months. Short version of that troubleshooting:
- purchased in October, the day they became available on Newegg
- ran beautifully until December, when I moved from XP to 7; constant crashes under 7 in both 2D and 3D (yes, I did a fresh install and then a fresh reinstall)
- I finally discovered that the 9.8A drivers (September Catalyst set, also shipped with the card) were rock-solid - no GSODs in games or other errors in 2D - but all later drivers would lead to a GSOD within 5-30 minutes
- XFX found no issues with my old 5870 in their labs, but agreed to send a replacement anyway

My replacement card still exhibits these errors, but much less so. I now get a GSOD every 45 minutes-3 hours with current drivers (and the 9.8A drivers are still solid). Things have definitely improved, in other words - it's not just placebo effect - but the same problem occurs and it's still not acceptable.

I've already tried:
- underclocking the card, both GPU and RAM
- increasing the fan speed
- reinstalling, reseating, etc.
- updating chipset drivers and BIOS
- experimenting with different memory timings and voltage settings (GSOD frequency seems lower when I overvolt my RAM, but that might just be voodoo)

Nothing is overclocked, aside from my experimentation with different memory settings. My system runs memtest and prime overnight with no issues. As I mentioned, the 9.8A drivers work fine; I only see problems with later drivers. Furthermore, my 8800 GTX worked perfectly for the 3 weeks or so during which I waited on XFX's RMA process - not a single crash or other issue.

All this leads me to think that, if the AMD card is finding a problem with my system, it's a problem which doesn't seem to be an issue for any other piece of hardware I own.

Any suggestions? My system is rock-solid from the perspective of memtest, prime, etc., so I'm not sure how to further diagnose this issue.

Giga P55 UD3R / i5 750
4GB g.skill Ripjaws Series DDR3 1600 (PC3 12800) @ 8-8-8-24 (the default advertised settings)
HT Omega Claro soundcard
Windows 7 Pro 64-bit
 
Last edited:
If you are running any third party utilities for overclocking the card or creating game profiles, try removing them from your system.
 
If you're running any third-party monitoring software at all, close it. Somehow they conflict with one another.

I did, and my constant problems the previous six months vanished that very moment. Haven't had a GSOD in almost a month, and right now my system's been up for 17 days without a problem.

I was never convinced it was my card, and indeed, it wasn't... I even put it back to its manuf OC of 875/1300...
 
I have been dealing with a similar issue with the XFX 5870 xxx edition since 3/15/2010.
I have had three cards total, each with related issues.

You might try creating a profile and setting the 3D clock settings for all 3 powerstates to see if that changes the frequency of the crashes (keep in mind after each restart/crash the profile will have to be re-enabled).

This seemed to be the only thing that remotely made a difference, although not a fix.
 
Tolyngee said:
If you're running any third-party monitoring software at all, close it. Somehow they conflict with one another.

I did, and my constant problems the previous six months vanished that very moment. Haven't had a GSOD in almost a month, and right now my system's been up for 17 days without a problem.
Click to expand...

That's interesting and a little alarming. I do run SpeedFan, so I'll try disabling it and see whether that produces any change. I only started running SpeedFan after the problems began, tho, so I'm guessing/hoping it can't be a factor.

Tolyngee, would you be willing to share what monitoring software you were using which conflicted with the card's operation? Don't worry about it if you'd rather not, but I'm really curious.
 
Last edited:
suibhne said:
That's interesting and a little alarming. I do run SpeedFan, so I'll try disabling it and see whether that produces any change. I only started running SpeedFan after the problems began, tho, so I'm guessing/hoping it can't be a factor.

Tolyngee, would you be willing to share what monitoring software you were using which conflicted with the card's operation? Don't worry about it if you'd rather not, but I'm really curious.
Click to expand...

Haha, "after" sounds like me...

Initially, with the 9.12 or 10.1 drivers, my problem was the desktop was very unstable (multiple GSODs a day), but games were fine as I was playing games after the Steam Holiday sales without any complaints.

After 10.2 or 10.3, my system flip-flopped. Desktop was almost perfectly fine, but games were now the problem. Made me not point at the card as bad, as with a driver release the problem completely flip-flopped.

Well, sometime around there, I was running MSI Afterburner, CPU-Z, Everest, and sometimes CCC Overdrive thingy, or whatever it is called.

Apparently any two of those used simultaneously will cause some odd issue. Simply put, the easiest way to spot the problem is that at the desktop, when your core clock should be 400, and your GPU voltage should be 1.062, multiple programs monitoring the card for some reason causes the card to go to a voltage of 1.649(? if I rem right), which is OCing the voltage well beyond what even any OCing tool would ever let you do (MSI:AB maxes at 1.350, well below 1.649)

Hadn't used CCC in quite some time, but the first three things I would open when re-booting Windows was MSIAB, Everest and CPU-Z so I could sit there any over-monitor my system to see what on Earth was going on. And this was causing the problem.

Now I boot... and just load MSIAB... Card GPU volts stay at 1.062, and problem had instantly disappeared.

Do us both a favor. Download MSIAB, and just run it. Don't start uninstalling or stop running anything just yet. I'm just curious if your GPU volts are way off like mine were.

If they are, that's your problem. Use MSIAB to determine what you cannot have running, as MSIAB alone won't cause the problem, but is the easiest way I know of to monitor your GPU volts.

Also, if you GPU volts go to 1.649 at the desktop, they'll appear to be normal inside a game (1.125), but in my experience your fan won't work right... Sometimes my fan would completely stop and the card would very quickly start to overheat... alt-tabbing to the desktop would get the fan working again when clock and volts changed, but would quickly stop working again when I alt-tabbed back into the game.

But also, if your GPU volts go to 1.649, closing anything won't fix that... You need to reboot to reset it... But again, when you reboot, use MSIAB to determine if anything causes it to go to 1.649... Also, it may take time for your card to finally go to 1.649, since it's programs conflicting that causes it... The quickest mine would go from 1.062 to 1.649 was 30 min... So, unfortunately no, rebooting won't instantly cause the problem to re-appear. So, don't just reboot, see 1.062 and think the problem is resolved. Monitor with MSIAB... If 24 hours after a reboot (I leave my systems on 24/7) your desktop is still at 1.062, then you've fixed the problem.

If this problem doesn't exist for you at all, then I apologize for your time.

But, it was 100% my problem, it lasted for a long time as I was monitoring anything and everything trying to spot a pattern to my problems, and closing all but one monitoring program (along with a reboot, of course), was an instant fix for all of my problems...

My computer's now been up 17 days... For the previous six months, that level of stability was just a pipe dream...

(And yes, I am willing to bet Speedfan and CCC Overdrive or Speedfan and MSIAB together, will cause the problem... Just run MSIAB...)

If your card's running right, you won't need to adjust anything ever. I just run MSIAB now to monitor my framerate. I run it for no other reason now.
 
It seems that people who get it usually will get it with the next card also. Sounds that it may be some other hardware that causes it but only with HD58xxx cards.

The chance of getting three duds in a row is just too low. Try the card with another motherboard.
 
tannat said:
It seems that people who get it usually will get it with the next card also. Sounds that it may be some other hardware that causes it but only with HD58xxx cards.

The chance of getting three duds in a row is just too low. Try the card with another motherboard.
Click to expand...

If the reason is what I cited, you'll get it with every card though, when it's not the card at all. He's already admitted to running CCC and Speedfan...
 
Had the problem with launch card from Oct 09. Bought another HD5870 last week for testing - within 24 hours it GSOD on me. I'm convinced its a hardware issue they will never resolve now.
 
dnottis said:
Had the problem with launch card from Oct 09. Bought another HD5870 last week for testing - within 24 hours it GSOD on me. I'm convinced its a hardware issue they will never resolve now.
Click to expand...

Until a month ago, I was so convinced of that fact that I even bought a GTX 480 (been sitting next to me for two weeks now) to replace the 5870.

Without tweaking anything, without new drivers, my problem instantly disappeared once I stopped loading Everest & CPU-Z, and only used MSIAB. Instant rock-solid stable system.

So solid, I even re-OCed the CPU. That was three weeks ago. Rock-solid. There goes another theory that the cards are sensitive to CPU OCs.

FWIW - six weeks ago (two weeks prior to a rock-solid system) I was experiencing sometimes six GSODs in a day. That's why I bought the GTX 480. Then I noticed my GPU volts, posted that info, someone told me what the hell was actually wrong, and viola! Instant rock-solid stable system.

And yeah, sometimes I couldn't game five minutes without a major crap-out, just like the OP. But insta-fix worked for me.

Look, the OP has already stated he's ran the CCC and some fan software at the same time. Stuff like that is what caused my headaches. It's very much worth it for him to look at that angle and determine yes/no regarding it than disregard it as a possibility.
 
Hmm, do you think this would extend to the stupid system resource widgets, etc?

I'm having a hell of a time diagnosing a 5770 that constantly GSOD's, and I never thought that the hardware utilization widgets etc could be causing it...
 
DarkStryke said:
Hmm, do you think this would extend to the stupid system resource widgets, etc?

I'm having a hell of a time diagnosing a 5770 that constantly GSOD's, and I never thought that the hardware utilization widgets etc could be causing it...
Click to expand...

If it is polling the hardware, then maybe.

Forgive my ignorance, by I'm not certain what a widget is. (If I'm stupid, I'm sorry, but I'm trying to help. I'll Google widgets in a moment...)

My best advice is download MSIAB and use it to monitor. If it's showing 1.649 on the GPU volts, then something besides just MSIAB is polling the GPU and is potentially causing problems. Keeping my GPU from going to 1.649, regardless of cause, is what was the instant fix for my system.

(That 1.649 might only apply to the 5870, not sure about other cards, come to think of it. But, I'd imagine if you see different GPU volts on boot that change later for no reason, then that's a problem, and determine if something's over-polling your GPU. Your GPU volts at the desktop should remain the same day in and day out at the desktop. (Naturally when your core clocks change for gaming your GPU volts should adjust, but your desktop volts at boot when you first load MSIAB with nothing else monitoring should be where your GPU volts stay at desktop the entire time.))

Hope that made sense?
 
I mean that exact thing. I always just thought they're annoying but one of my client builds loves having that misc crap running. It would explain why I never ran into those problems when I tested the machine countless times for hours (prime 95 + furmark running for hours on end), yet when they use it the machine greyscreens pretty quickly..
 
DarkStryke said:
I mean that exact thing. I always just thought they're annoying but one of my client builds loves having that misc crap running. It would explain why I never ran into those problems when I tested the machine countless times for hours (prime 95 + furmark running for hours on end), yet when they use it the machine greyscreens pretty quickly..
Click to expand...

I bet if you load MSIAB at boot and take note of the GPU voltage, that after running it and that GPU observer that you'll see the GPU voltage increase at some point due to both polling the GPU...

Like I said, instant fix for me... Your experience with their machine vs. their experience with same machine sounds like the same story... System's fine, 'til you start running monitor tools and begin over-monitoring the system...

I can't prove it from here, but if it just happens when he games and happens pretty quickly, I am willing to be from my own experience that it doesn't take long for the fan to stop, he doesn't realize it's stopped (I was running mine for no good reason at 100% (loud loud loud) and it was obvious when it would stop because it got quiet), but he doesn't realize it has stopped, it quickly overheats and GSODs...

That fan stopping and card overheating on my 5870 would happen very quickly... Usually the fan would just slowly stop... 70%... 60%... 50... 40... at which point it's already slowly overheating and at almost 100c very quickly...

I didn't catch on to my GPU volts going bonkers 'til I booted one time and the volts were at 1.349... Ever since then I've never been able to get it to go to 1.349, so no idea how it got there, certainly couldn't have been my own doing... But that's I started paying attn to the GPU volts, and realized it was doing some bizarre stuff...

MY GPU would either GSOD, or attempt to underclock the core down to 600 then even 400 to compensate for the fan not working... This lead to poor framerates...


Well, sorry for all the babble...

But yes, closing all monitoring software except for MSIAB and a reboot was an instant fix... And I absolutely was convinced ATi drivers or something was crap and was going to install my GTX 480 and throw the 5870 in the garbage...

I'm going to go with the GTX 480 now though... THe only reason I still have the 5870 in the machine is I want to see just how stable my machine is now... It's now been up 18 days... So, I think I'm going to call it 100% fixed and the card perfectly fine and call it a day and finally take the card out...
 
Tolyngee said:
I bet if you load MSIAB at boot and take note of the GPU voltage, that after running it and that GPU observer that you'll see the GPU voltage increase at some point due to both polling the GPU...

Like I said, instant fix for me... Your experience with their machine vs. their experience with same machine sounds like the same story... System's fine, 'til you start running monitor tools and begin over-monitoring the system...

I can't prove it from here, but if it just happens when he games and happens pretty quickly, I am willing to be from my own experience that it doesn't take long for the fan to stop, he doesn't realize it's stopped (I was running mine for no good reason at 100% (loud loud loud) and it was obvious when it would stop because it got quiet), but he doesn't realize it has stopped, it quickly overheats and GSODs...

That fan stopping and card overheating on my 5870 would happen very quickly... Usually the fan would just slowly stop... 70%... 60%... 50... 40... at which point it's already slowly overheating and at almost 100c very quickly...

I didn't catch on to my GPU volts going bonkers 'til I booted one time and the volts were at 1.349... Ever since then I've never been able to get it to go to 1.349, so no idea how it got there, certainly couldn't have been my own doing... But that's I started paying attn to the GPU volts, and realized it was doing some bizarre stuff...

MY GPU would either GSOD, or attempt to underclock the core down to 600 then even 400 to compensate for the fan not working... This lead to poor framerates...


Well, sorry for all the babble...

But yes, closing all monitoring software except for MSIAB and a reboot was an instant fix... And I absolutely was convinced ATi drivers or something was crap and was going to install my GTX 480 and throw the 5870 in the garbage...

I'm going to go with the GTX 480 now though... THe only reason I still have the 5870 in the machine is I want to see just how stable my machine is now... It's now been up 18 days... So, I think I'm going to call it 100% fixed and the card perfectly fine and call it a day and finally take the card out...
Click to expand...

I still get GSODs when running only MSIAB and furmark. :/ Ony tried 10.4 so far
 
davyp3 said:
I still get GSODs when running only MSIAB and furmark. :/ Ony tried 10.4 so far
Click to expand...

You're using MSIAB to determine if your GPU volts are over-volting.

The only real way to test this is to let it sit at the desktop and see if the GPU volts change when the GPU's not being stressed.

But MSIAB showed nothing unusual while Furmark was running? When being stressed, MSIAB will still show the volt staying at 1.125 (where they should be for a 5870), but the temps would go up even though the fan speed is supposedly going up. Then the clock would start underclocking itself, while the fan speed starts dropping but the temps keep going up...

This continued 'til I expected it overheated itself and GSODed...


When Furmark's running, every one in a while stop it and let everything be idle and just see if your GPU volts at idle are where they were when you first booted.

If everything's looking like it should, and you are still having GSODs, then perhaps unfortunately your issue isn't what was my issue...

But mine was an instant fix, so I'd hate to not be vocal to others about the possibility... (because if it is the issue, a billion RMAs won't solve it, and you're just wasting your own time and $$ with the RMAs...)
 
mine finnally fixed whn reducing my oc from 4.3 to 4.25 on my cpu and ram follwed suit a bit..
fixed...my gsod's
 
Well, I've confirmed it's not, for me, tied to any monitoring software. I never used that stuff all the time anyway, and my GSODs persist even when they're all unloaded - i.e., whether I used just MSIAB, or nothing at all. But thanks for the suggestion, Tolyngee.

However...I think I may have found the actual culprit, based on reading another thread here. My G.SKILL RAM (F3-12800CL8D: 8-8-8-24 @ 1600) has always passed overnight memtest runs with flying colors, and my 8800 GTX had no problems with it - but another thread here convinced me to reject my assumption that "memtest pass = bombproof RAM". I tried dropping the multiplier on my RAM, and bingo, my GSODs went away. I've now played BC2 for about 5 hours at the new multiplier without any crash, even tho it gray-screens every 30 minutes @ 1600.

I won't feel totally confident in this "solution" until I spend more time in-game and continue to confirm the lack of GSODs, but I've started a troubleshooting thread over at the G.SKILL forums to get more guidance on settings to test.

It is disconcerting to think that the 5870 might be picky about RAM performance that produces no errors for other videocards, memtest, and my whole system...but it's a relief to finally discover something that seems like progress. I just hope I'm not back here in a day or two saying "never mind". :D
 
I had a lot of issues getting 10.6 installed. Thought I had things fixed with the GSOD. I'm firing up a stupid Pop Cap game (so simple) and my whole box is freezing. I tried underclocking, overvolting. Memtest is doing fine. I went all the way back to the 10.3 cats (non beta) and Pop Cap runs fine now (ie: getting farther in the game than ever before).

Just to be clear, I express uninstalled 10.6. I booted to safe mode and removed the devices. Rebooted. Installed 10.3. I'll update if something else starts flaking out.
 
Try bumping up the voltage of your northbridge chipset by 0.1v. Let me know if this helps.
 
suibhne said:
But thanks for the suggestion, Tolyngee.
Click to expand...

Bummer. I'm sorry for taking up your time chasing what wasn't your problem.

But, it was worth a mention. That was 100% my issue, and it was an instant fix to resolve... Not a single GSOD since that moment I stopped loading extra monitoring software.
 
My problems tend to relate to voltage. I've gotten GSODs when I'm pushing my cards extra hard (higher than 1015MHz+). Raising my northbridge vcore resolved this.
 
suibhne said:
However...I think I may have found the actual culprit, based on reading another thread here. My G.SKILL RAM (F3-12800CL8D: 8-8-8-24 @ 1600) has always passed overnight memtest runs with flying colors, and my 8800 GTX had no problems with it - but another thread here convinced me to reject my assumption that "memtest pass = bombproof RAM". I tried dropping the multiplier on my RAM, and bingo, my GSODs went away. I've now played BC2 for about 5 hours at the new multiplier without any crash, even tho it gray-screens every 30 minutes @ 1600.
Click to expand...


I had this exact same issue with 2gb 5870s. Dropping ram by one multiplier fixed it. No idea why, system was very stable with the 1gbs in there, but not with the 2s, 30mins of gaming =GSOD. I originally thought i needed more Vcore, but no change. Also tried more and less Vqpi. No change there. I tried all kinds of things actually. But for me dropping the ram back a notch below 1600 fixed the issue. I though it was my mobo that wasn't playing nice with the cards but now i see someone else with a similar fix, hmm. Wish i'd got a pair of 480s instead... :rolleyes:
 
You must log in or register to reply here.
Back
Top