Possible problem HD 5870

[andmar]

I bought an entirely new setup at the end of last year consisting of ...

Antec 1200 tower case
intel core i7 930
gigabte ga x58 ud3r rev2
corsair h70
corsair hx850
corsair 2000mhz 6gb
xfx hd 5870 radeon xxx edition
Windows 7 64bit Ultimate

And I've had no end of problems with BSOD and GSOD (Grey/Green stripes of death, basically you get vertical stripe lines and your computer completey hard locks, you can not even ctrl-alt-del!) crashes and it's driving me insane, I've tried so many things to try and make some kind of difference, so far they consist of:

*Flashed the card to the latest bios on the xfx support site
*Installed and tried every ati driver release from the xfx site from 10.10 onwards and used driver sweeper on each occasion to make sure all remnants of the old drivers were removed.
*Memtestx86'd passes 100% after 1 full pass
*Removed all traces of gigabyte DES2 energy saving and it's realted grdv.sys driver that can apparently causes crashes on gigabyte motherboards with setups requiring a lot of power
*Running Microsoft driver verifier
*Updating all my motherboard drivers
*I'm running my harddrive in AHCI mode not RAID or IDE
*My memory/cpu is running at totally normal, not over clocked settings

What I have noticed is:

*Only intensive games like half life 2, left 4 dead 1/2, civilisation V, settlers 7, battlefield 2 bad company. gta iv (i.e. not older games) and benchmarking like furmark seem to make it crash
*BSOD's with memory.dmp dump files always return...

Unable to load image \SystemRoot\System32\drivers\dxgkrnl.sys, Win32 error 0n2 BugCheck 116, {fffffa80057164e0, fffff8800492e008, 0, 2}
Probably caused by : atikmpag.sys ( atikmpag+7008 )
Followup: MachineOwner ---------

I have a log of crash dumps if anyone wants more info.
*GSOD's with a complete hard lock as well as artifacts and GSOD's that recover momentarily before later fully hard locking, black screens for 10-15 seconds then a recovery.... all of which can happen anywhere between 10 minutes or 4-6 hours!
*The temperatures in the case are not much higher than ambient, and the card gets to 60-70 at the very most. The case and cpu together have 8 fans!
*Turning my virus checker off seems to "help" (i.e. prolong the length of operation) but not remove the problem (i.e. it still eventually crashes!)
*The stripped crash (GSOD) has never occured when "not" in games or benchmarking, so just fiddling around on the internet is fine.

I'm now at the point of considering asking for an RMA from scan, but I'd like to know if anyone can think of anything else I could try?

[Jasp]

Bearing in mind the XXX is a overclocked card knock it back down to factory clockspeeds in CCC. GPU should be 850mhz, memory should be 4800mhz and see if the problem still occurs.

[Phage]

Still no luck ?
I know you've tried everything. Get an RMA.

[andmar]

Hey phage, still floating about, still no luck!

One interesting thing that I thought I might throw everyone's way that I noticed last night whilst testing with the latest furmark, I was about 2-3 minutes into running a stability test on the latest (on xfx's website) 11.1 drivers whlist running logging on the card`s temps/voltages/clock speeds etc, and then artifacts began to appear then there was a brief blip of a second or so of the grey stripe crash then it recovered, that`s nothing unusual, but, what I did notice was in the logs it clearly showed that the card temperature dropped by roughly 5C during the period of the mini stripe lock up, very weird!

Has anyone seen that before?

Jasp - as phage, suggests I have tried "a lot" of stuff! see here... http://forums.hexus.net/graphics-car...d-issue-2.html
I did try "underclocking" the card (before putting it back to manufacturers settings), and this did seem make it work stably (although I didn't test it for a huge amount of time), but I've paid extra for an overclock card so it should do what it says on the tin.

I'm just interested if anyone on here or the scan guys can perhaps offer any advice on things I should try testing or perhaps even stuff I should check to do with memory/motherboard... etc. I'm not entirely certain my memory is right, at 533mhz the motherboard seems to be setting the timings slightly differently to the SPD (which I've tried manually setting to what the SPD says) and at 666mhz it looks correct as the motherboard sets it (all according to CPU-Z), either way you "still" get the "exact" same crashes... which is why I "think" I have all that kinda stuff correct.

If I really can't find anything then I want to RMA, I am trying to rule everything out as I don't have any access to another pci-e x16 computer so I can't do the obvious "try it in another computer" test...

p.s. oh and also, after extensive conversation with the support at XFX they seem to think I should send it back to scan!

[Phage]

I agree with XFX. You'll need an RMA number from them. They should have given you one.
Then get straight on the blower to Scan.

[andmar]

I've asked xfx... it's weird as they just send "sounds like you should send it back to your reseller", but didn't offer an RMA number?

Waiting to see what they come back with

[andmar]

Just got my xfx rma and I've been straight on the phone to scan. Let's hope it's all a quick turn around *crosses fingers*

[Phage]

Good luck. Sorry I wasn't of more help.

[EzyRyder]

It sounds like a faulty card mate. RMA.

[andmar]

In case anyone might be still out there floating and still interested, this saga is starting to get tiresome... I'm now in the position that my card has been RMA'd and is (presumably still) currently at XFX, they have apparently found no fault which I simply find defying belief! I am now currently in the position of waiting on their (somewhat slow) ticket system for a tech to get back to me and actually tell me what they did, what hardware combo they used, how long they tested it, I'm amazed they got that card to work if they have indeed tested "thoroughly"! I can completely understand if they shoved it in a pci-e slot, turned the box on and went "tick, it's working", yes, it will do that, but pushing it in furmark for long enough or playing a game for longer than an hour/2hours or so and... nope, artifacts/GSOD/BSOD! grr

The last correspondence I had off them was basically stating the card is fine and I should accept it back from scan. Well that's amazing!

I've been looking around on the net since and I'm starting to notice a correlation between my problems and others woes surrounding XFX and their RMA process. It sounds like they trade on the good nature of their american RMA process but the UK/European customer experience appears to be quite different.

[Chris P]

andmar

Please can you PM me your invoice number and I will look into this for you.

Best Regards

[andmar]

Popped a reply off to Chris, I just wish I'd seen his reply sooner!

I do hope Scan can help me with this!

[Chris P]

andmar

We have e-mailed XFX directly, pointing to this thread and to explain the test process your cards have gone through to determine if this was inadequate.

Best Regards

[go.mole]

Hi Andmar,

I'm suffering from the same problem, Scan kindly sent mine off for me, but XFX didn’t done anything to it and as you can see from this thread (http://forums.hexus.net/scan-care-he...-238533-a.html) Scan haven't got back to me to give me any information from XFX

I'm still in the middle of trying different things to get it running at XXX speeds, it seems to work underclocked (stock 5870 speeds) but as with yourself, if you pay for a factory overclocked card you expect it to work.

I’ll be watching how this plays out for you as my saga goes a long way back. Scan have at one point had my whole computer and told me they didn't have a single crash in ~36hrs of it running benchmark software, so therefore no faulted it. I think you probably need to take the whole computer to Scan and play some games at their support desk so they can see that in fact it does crash!

If you can prove that you do have a problem, then I may have hope that I may get the problem sorted as well.

Mole

[andmar]

Excellent stuff Chris! I just had a ticket from XFX saying they are accepting the card back and will be testing again for an extended period. This is good news. If they can test it for a prolonged period (5-6 hours or more) and still show no artifacts, no blank screens, no GSOD's (that recover or not) and no BSOD's then I'm slightly more willing to beleive that it is a compatibility problem, even so, I am not running any out of the ordinary components and I have been running them at entirely standard settings, I have nothing overclocked (other than the graphics card which is meant to be as it's an XXX edition)

My only concern is that you can not perform a simple "end to end" test whereby you leave the computer running and return to see if it has crashed, that IS NOT sufficient as very often the card works and does not crash but is still doing odd stuff (artifacts, blank screens)

go.mole, what is your set up? As exact as you can remember, I'm wondering if we have any components that correlate?

[Chris P]

go.mole

I am in the process of looking into your particular situation.

Best Regards