ATI 5850 / 5870 GSOD & Corruption Poll

Hi Guys,

After much reading here and elsewhere I wanted to try and get a clearer picture of the graphical corruption and screen of death issues a lot of us are having with our 5850 and 5870 cards (I know the 5890s are having lost of issues too but I think this card should have a different poll because it is a twin gpu card).

Specifically I would like to know what OS you guys are running when you get issues or whether everything is running fine.

I realise there are 2 threads running already about this but I felt a poll thread was appropriate too as this issue seems so widespread.

You can select more than one choice to reflect your testing.

5850 & 5870 only (if there is high demand I will happily start a similar poll for the 5970).

Anyway, vote away folks and lets put some numbers to this issue :)

GSOD & VGA Driver recoveries, 5870, all Windows OSes, overclocked & stock.
System is 100% stable, needless to say.

Quote:

---

Originally Posted by BenchZowner

GSOD & VGA Driver recoveries, 5870, all Windows OSes, overclocked & stock.
System is 100% stable, needless to say.

---

Dammit I was hoping to be safe with XP 32 bit ;)

Definitely looking like hardware in your case...

I take it that this was more than one card / brand?

Were the OS's patched / unpatched or did it not seem to matter?

2 different cards, both showcasing the same symptoms with all drivers ( 9.10, 9.11, 9.12, 9.12hf, 10.1, 10.2b )

XP SP3.
Vista pre-SP1, Vista SP1, same result
Win 7 same updated & not updated ( OEM disc )

Asus & Jiggabyte :p:

Running a reference Sapphire 5850 on Win 7 64bit. After 10.1 GSOD hotfix driver I didn't get another GSOD for over 10days ... than I flashed the new sapphire BIOS (keeping same driver), no GSOD for another few days. Seems to be fixed for me.

Never had any GSODs in 2D. Not with the hotfix drivers, nor the drivers before those. 3D is a completely different story, however. GSODs, striped screens, solid colored screens, driver recoveries, you name it. No matter what driver/bios. It happens both at stock as on overclock settings, even though the overclock settings are 25 LinX loops, 24h blend and 1h OCCT stable. Tried another 30 loops of LinX to see if anything degraded, but it was still 100% stable. The videocard passes multiple hours of OCCT GPU test at shader complexity of "3". I'm completely at a loss here. I even went as far as running LinX and OCCT GPU at the same time to see if higher power usage made things unstable...It didn't. When switching drivers I always make sure I completely uninstalled the previous driver and even run driversweeper for good measure. Happening on Vista x64

i picked the last option, this 5850 runs 850/1200 24/7 and i never get any of this, my preferred driver is ATI_8.70_RC2_Vista_Win7_Jan14. its the best since 9.12 hotfix for me and i have used EVERY beta out there.
I use the msi 5850 bios and its a dream, i like the card better than the 5870 i had.

Quote:

---

Originally Posted by Rattle

i picked the last option, this 5850 runs 850/1200 24/7 and i never get any of this, my preferred driver is ATI_8.70_RC2_Vista_Win7_Jan14. its the best since 9.12 hotfix for me and i have used EVERY beta out there.
I use the msi 5850 bios and its a dream, i like the card better than the 5870 i had.

---

Why did you switch from a 5870 to a 5850? And why do you like the 5850 better? I mean, sure, the 5850 might be a better videocard from a financial perspective, but the 5870 is simply the fastest of the two. Is your 5850 more stable or something?

I didnt have stability issues with the 5870 really, only real sweet driver was 9.12 hotfix for it though, I game at 1920x1200 at 4x AA and the 5850 with a decent OC isnt really any slower than a 5870 that I can tell on a 775 setup or 1156 setup, both overclocked of course.
Plus, I just like to play with video cards :P

Pair of XFX 5850's in crossfire here and havn't had a single issue in the 4 weeks I'v owned them. (touch wood)

Ok.

Had my 5870 for about 8 hours and no 3D corruption or GSOD's yet.. but....

I'm getting some interesting stuff in 2D:

http://i263.photobucket.com/albums/i...mmm/58702D.jpg

:eek:

Windows XP SP3 x86

10.2 beta (full install)

I voted Xp 32b, Vista 32b, 7 32b and 7 64b. All the same crap.

A good majority appear to have probs in win 7 64bit

Quote:

---

Originally Posted by jaredpace

A good majority appear to have probs in win 7 64bit

---

Because most of them have tested their cards in Win7 x64 only :p:

Quote:

---

Originally Posted by BenchZowner

Because most of them have tested their cards in Win7 x64 only :p:

---

My thinking precisely :yepp:

My two are fine, but if there was a problem I'd have to report it as Windows 7 64bit as that's what I use exclusively. As I suspect do a lot of the people that voted so not much point.

Issues? What Issues? My card is running fine! XFX 5870 version 1.5 drivers are 10.1......... Vista/64

Quote:

---

Originally Posted by fornowagain

My two are fine, but if there was a problem I'd have to report it as Windows 7 64bit as that's what I use exclusively. As I suspect do a lot of the people that voted so not much point.

---

Which OS has the most corruption / GSOD's is not the issue here.

What this poll is attempting to establish is whether these issues occur independently of what OS is used or if there might be some causal link.

A lot of people will test on one OS only.

Others will test many.

All votes are still valid.

From my personal testing, the issue has absolutely nothing to do with the OS, 32b or 64b. As much as AMD would like to make us think this is our problem, it's 100% theirs.

My card was GSODing in taxing titles like Dirt2 or Cysis when memory was OCed or stock.
But with driver fix I've tried stock clocks and nothing is happening. Seems that my issue was sorted by driver update then :)

PS. just to be clear it did happen couple of times on the desktop as well, but I had it maybe once in a week time or so.

Problem with this poll is that the use of different OS is not spread out evenly. If we assume that the problem appears on all systems, the only result we can conclude is that half the people don't have a problem, and 1/3 of the people with problems run 7 64 bit. This does not neccesarily mean 7 64 bit is more sensitive to the problem than other OS, it could also just mean there are a lot of people running windows 7 here :).
To get conclusive results, we'll need another poll in the general forum asking everyone what operating system they use (if it doesn't exist already). Combining both numbers should give an estimation of each OS's sensitivity to the problem.

Quote:

---

Originally Posted by Biker

Which OS has the most corruption / GSOD's is not the issue here.

What this poll is attempting to establish is whether these issues occur independently of what OS is used or if there might be some causal link.

A lot of people will test on one OS only.

Others will test many.

All votes are still valid.

---

Nope its going to be massively biased, its only relevant if all the OS's were tested on the same cards. It could still be a W7 x64 problem or not, you can't tell from that poll. You'd need an equal sample size from each OS to balance and that's obviously not going to happen in an open poll. Best that will tell you is the ratio of people bothered enough to answer with problems vs those who haven't.

@fornowagain...
I tested 2 cards on every OS ( Win XP x86/x64, Win Vista x86/x64, Win7 x86/x64 )...
Same :banana::banana::banana::banana:e.
And I'm quite sure if somebody from the Win7 x64 list decides to install another OS he'll have the same symptoms as well ;)

I don't doubt it, doesn't validate the poll though.

Now like BenchZowner everyone has to test and add a vote for every OS listed... ;)

----------------------

@BenchZowner, interesting you have two cards that do it. Have you tried either in another system? With a different motherboard + psu etc would be nice.

Quote:

---

Originally Posted by fornowagain

Nope its going to be massively biased, its only relevant if all the OS's were tested on the same cards. It could still be a W7 x64 problem or not, you can't tell from that poll. You'd need an equal sample size from each OS to balance and that's obviously not going to happen in an open poll. Best that will tell you is the ratio of people bothered enough to answer with problems vs those who haven't.

---

Look.

The point of the poll is to find out how many users have issues.

The point of the poll is not to narrow it down by operating system.

If the poll is not set up empirically or objectively enough for your exacting scientific standards then I can only humbly apologise.

It would have been impossible to have set the poll up in the more open ended way you suggested (whether more desirable to do so or not) due to the restrictions on poll construction here by Vbulletin and Xtremesystems.

At the moment it is running at pretty much 50% having issues and 50% not having issues.

That is the interesting data here.

Quote:

---

Originally Posted by Biker

Look.

The point of the poll is not to narrow it down by operating system.

---

No you look, then don't put OS's up then, not unless people have tested against all of them. Its utterly pointless, the way it is a simple yes or no would be just as accurate. The sarcasm is undeserved so climb down from your horse its not helpful.

I've never had any issues but maybe cause I've been using the new BIOS right from the start.

Quote:

---

Originally Posted by fornowagain

No you look, then don't put OS's up then, not unless people have tested against all of them. Its utterly pointless, the sarcasm is undeserved so climb down from your horse its not helpful.

---

Well if you have nothing useful to offer and / or are not suffering these issues as a lot of us are then please go and find a different thread to troll in.

Edit:

10.2 Drivers now available from here

Edit 2

10.2 Windows XP 32 bit @ Megaupload

10.2 Windows XP 32 bit @ Mediafire

Edit 3:

10.2 Vista & Win 7 x64 @ Megaupload

10.2 Vista & Win 7 x64 @ Mediafire

Quote:

---

Originally Posted by Biker

Well if you have nothing useful to offer and / or are not suffering these issues as a lot of us are then please go and find a different thread to troll in.

Edit:

10.2 Drivers now available from here

---

I don't troll. If you bothered reading my comments, I already have asked a pertinent question. I also have two of these cards overclocked from the day of release with early bios' and no problems. You're way too defensive. It's just as important to determine hardware, settings etc from working systems as it is from those with problems.

Quote:

---

Originally Posted by fornowagain

@BenchZowner, interesting you have two cards that do it. Have you tried either in another system? With a different motherboard + psu etc would be nice.

---

Both cards tested on 2 different ( 100% stable though, also tried @ stock to be 100% sure ) systems...

System 1 - Gigabyte GA-EX58-UD5 + i7 920 - 6GB Chaintech DDR3 - Corsair HX620
System 2 - eVGA X58 Classified + i7 950 - 6GB A-data DDR3 - modified OCZ ProXStream 1000W

Tested both cards with stock BIOS & MSI's unlocked BIOS from Garrett Wu.

And both card mucked around on both systems? Exactly the same issues?

For reference, both mine are on the Asus bios, originally on Sapphire. On a Gigibyte x58 Extreme, pretty much the UD5 board. PCP&C psu.

Same issues ( GSOD @ 3D, and sporadic driver recoveries ).
They're out of control, and cannot be replicated as easily as someone would think.
The funniest part of this crazy issue is that I can get a GSOD at stock clocks easier and more often than at overclocked frequencies :D

I have yet to experience any 2D corruption issues at all, but some other members and people I know have :D

What boggles me is how this can be a driver issue since all of us use the same drivers and the same OSs yet some people are fine and some are badly screwed...

Programming device support software (drivers) is probably the worst job any programmer could have, because there are an infinite number of variables involved (Or so my professor told me). I believe this situation has happened before where some where affected by a problem, and some weren't, although they were using the same drivers.

A recent example would be the corrupted Intel firmware, where some people saw their entire SSD formatted, whilst others could run it without any problem.

I am having some fun with 3D corruption now :)

Tested with 9.12, 10.1 beta, 10.2 and 8.703 whatever that is ;)

Running @ stock clocks / volts:

3D Mark 01:

I am getting half the framerate for most of Nature (round about 420) which only jumps back to 999 briefly as the camera pans over the lake... anyone else getting this?


3D mark 05:

Corrupted smoke when the large canister drops on the minigun survivor at the end of Proxycon.

The smoke effect is replaced by very bright largish white squares which look a lot like the corrupted cursor that I also have had twice.

Also getting corruption during cpu tests 1 and 2:

In cpu test 1 as the camera pulls back I quickly start getting thn grey / brown lines appearing and dissapearing rapidly they are all displayed at isometric angles.

In cpu test 2 I also get the lines but these are joined by thousands of small black blotches that flicker on and off rapidly.



3D mark 06:

The smoke is also corrupted in proxycon in a very similar way to '05 except it is prettier / shinier :)

Both CPU tests are corrupted in exactly the same way as '05.



As an experiment I overclocked the card to 900 / 1300 and ran '06 again.

Unsurprisingly the corruption was worse:

In Proxycon wall textures were dropping in and out and the whole screen would flash.

In cpu 1 in addition to the '05 mess I also got for the first time vertical black / grey "prison cell" bars that many others have also been reporting.


In cpu test 2 I nearly had the whole screen flashing with large grey polygons in addition to all the rest ... I really thought I was going to get my first GSOD :D


This time around Canyon Flight was also not happy with the sea corrupted at the beginning, followed by the canyon walls then the monster and ship underbelly had that crappy mottled shadow effect texture very similar to the rubbish shadow effects in GTA IV.



After this I went back and repeated all tests 3 times and found that in my case the corruptions were easily repeatable.



Conclusion:

After lots of testing I am fairly cartain that no amount of updated drivers or hotfixes are going to get this card running well.

:yepp:

...

What I find hard to believe is that ATI basically has the solution at hand, but doesn't really use it.

First off, they release a hotfix for 9.12 which gets rid of the increased cursor. Works perfectly. But then, with 10.1, the problems return. Then they release a hotfix for 10.1 which gets rid of most of the GSOD's, which also works amazingly well. But when 10.2 is released, the amount of GSOD's start to increase again.

ATI, try combining those 2 hotfixes. You'll please a lot of people ;).

I have had lockup issues on the 5850 with the card overclocked with the MSI Unlock Bios when overclocked playing L4D 2 only. I don't know why LFD 2. Last time I played the card was screaming. There is no fan, since I watercool. The card itself was making noise. If I drop the OC on the card L4D 2 seems to play well w/o issues. Hopefully this will be fixed. All my other games run fine overclocked to 1300v, 1030clock, and 1200 ram.

I have the same problem. My 5850 gives off a high pitch noise when I'm using Google SketchUp :S. I'm pretty sure it's not the fan, but the card itself.

temp related issue maybe with VRM's or MEM IO...what are the MEM IO temps reading when errors happen..?

Voted "Win 7 x64" since i don't use anything else. I wonder if ATi can fix their cards with drivers or we need to wait for the new revised cards.

Quote:

---

Originally Posted by Alexandr0s

I have the same problem. My 5850 gives off a high pitch noise when I'm using Google SketchUp :S. I'm pretty sure it's not the fan, but the card itself.

---

The high pitch noise is a choke or coil "resonating" on the card. It most often occurs in 3D programs with very high FPS (dark screen at a game intro and such).

No issues with GSOD's with my 5870. It's a XFX card running win 7 x64. I've even been running under 100% load for over a week crunching Milkyway@home.

I get GSOD with catalyst 9.11 and 10.1 hotfix, (Win7-64)
and BSOD with 10.2 beta and 10.2 final. BSOD indicate a dxgmms1.sys corruption
I remembered that these problem appear after I updated my GA-EX58-UD3 bios to F8, F9 also.
Since I am back in the BIOS F6 (2 days), everything works perfectly!

=>BenchZowner: what bios you use for you Gigabyte GA-EX58-UD5? if it's F2 or F3, I suggest you to try to return to F1 for test...
Bios F2 & F3 provide same new feature that F8 & F9 for GA-EX58-UD3R (Add support for GulfTown and "Enhanced Memory capability ").

Interesting article posted over @ Toms covering 2D corruption & performance issues on the 5xxx series now opdated with responses from ATI / AMD...

Thankfully now also republished in English (Mein Deutsch ist nicht so gut, ja?) :D


Edit:

After a lot of reading and some fiddling around I have finally managed to get rid of 95% of my 2D corruption.

How you ask?

By disabling all cursor and advanced drawing graphics acceleration!


In XP: go to display properties > advanced > troubleshoot and then set hardware acceleration two notches away from full....



This is a great relief as trying to read a forum was nearly impossible as every time I scrolled up or down I instantly got half a page of unreadable garbage!


Note this does not get rid of it completely, to do that i need to completely disable hardware acceleration!

I haven't experienced it on either XP Pro 32bit or Vista Home Premium 64bit.