• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

How to know if a graphics card is faulty?

Associate
Joined
13 Jul 2011
Posts
16
Hi, I receieved the parts for my build yesterday morning.

Asus GeForce GTX 570 DirectCU II 1280MB GDDR5 PCI-Express Graphics Card
Intel Core i5-2500K 3.30GHz (Sandybridge) Socket LGA1155 Processor - OEM
Asrock Z68 Extreme4 Intel Z68 (Socket 1155) DDR3 Motherboard
Cooler Master CM690-II Lite Dominator Case & GX 750W Power Supply
Cooler Master GX 750W '80 Plus' Power Supply
Corsair XMS3 8GB (2x4GB) DDR3 PC3-12800C9 1600MHz Dual Channel Kit
Akasa AK-CCX-4002HP Venom CPU Cooler

Everything seemed to be working fine until I tried to run some benchmarks.
I tried to run 3Dmark 11 test demo, 3DMark vantage test demo, and also the unigine demos.

Almost everytime I run a benchmark It crashes with the "nvidia display driver stopped working and recovered". On the "Jane nash" test of Vantage, it seems to crash at exactly the same frame every time. But the odd thing is, on both 3Dmark 11 and unigine, I managed to get through a benchmark a couple of times, and then other times it would fail.

The error I get with 3DMark is
Workload Single init returned error message: eva::d3d11::rendering::scene_renderer::render(): draw_unshadowed_illumination_task for thread 0: File: device_context.cpp
Line: 515
Function: struct ID3D11CommandList *__cdecl eva::d3d11::deferred_device_context::do_finish_com mand_list(bool)

Expression: native()->FinishCommandList( restore_deferred_context_state, &result): DX11 call failed.

Device hung due to badly formed commands.
DXGI_ERROR_DEVICE_HUNG: ID3D11DeviceContext::FinishCommandList:

And the error I get with the Heaven 2.5 bencmark from unigine is:
"D3D11AppWindow::do_swap():device removed"

The errors seem to be random on 3DMark 11 and heaven 2.5.

Does this look like a faulty card? What else could cause this?

I would like to know what sort of tests OcUK would do to test for a fault, so I can do it myself and not waste money.

Also, I had the latest drivers installed.
 
Soldato
Joined
5 Feb 2009
Posts
15,910
Location
N. Ireland
is the gpu overclocked. a quick google of the above errors seems to indicate an unstable oc or not enough voltage to the card.

but some1 with more knowhow may be able to better advise
 
Associate
OP
Joined
13 Jul 2011
Posts
16
Well, apparently, the "stock" gtx570 has a core clock of 732Mhz. Asus factory overclocked the DirectCU II to 742Mhz. I haven't overclocked or changed voltage at all.
It currently uses 0.975v I think. The only thing different to the stock 570 is the core clock.
 
Permabanned
Joined
12 May 2011
Posts
1,119
Do you see any artifacts in games?

You could also try folding@home GPU client and see if you get any errors. A robust card should be able to do folding continuously for over 24 hours without any error.
 
Associate
OP
Joined
13 Jul 2011
Posts
16
Yes it has onboard VGA, But I don't think the onboard would be dx11 would it?
I don't know what folding is either.

I just found something out though. I ran dxdiag and under the display tab the ddriver of the card has a ddi value of 10.1.

Doesn't that mean the card can only support dx 10?
 
Soldato
Joined
17 Aug 2005
Posts
2,825
Location
SW Scotland
I got the following message in Heaven benchmark just the other day and it was because of an unstable OC (as commented above by someone else). If you GOOGLE this error message, this is also the consensus of opinion.

"D3D11AppWindow::do_swap():device removed"

Try increasing the voltage to the GPU a notch and re-test.

I should maybe comment that a lot of other benchmarks ran OK (including 30 minute run of OCCT). But Heaven would collapse in a heap (usually on the 2nd run through) with the above error and Crysis 2 would "black screen" after 5/10 minutes and Windows Event Viewer would show "... driver stopped responding". Increased voltage from 1 v to 1.024v and all was fine. Not necessarily your problem, but maybe worth a try, Download Afterburner and try.

There used to be a problem in W7 that this was by all accounts reported incorrectly by DXdiag but there was hotfix ages ago to sort this (or so GOOGLE says).
See http://support.microsoft.com/kb/975243
 
Last edited:
Associate
OP
Joined
13 Jul 2011
Posts
16
How much should I imcrease the voltage by though? I have no idea about that stuff :/

Also, did you see my reply about ddi version in dxdiag? If you have a DX11 card, can you check what your ddi version says under the display tab please?
 
Soldato
Joined
17 Aug 2005
Posts
2,825
Location
SW Scotland
Hi

Mine says "11"

So it almost sounds like you don't have the latest version of DirectX installed!

PS. "DDI Version' is the maximum DX level supported by hardware." So that sounds a little strange!
 
Last edited:
Associate
OP
Joined
13 Jul 2011
Posts
16
Well, I tried a fresh install with the drivers that came with it. The latest drivers, and new beta drivers that dont have whql.

I see you have an nvidia card, can you post a screenshot of your dxdiag?
 
Soldato
Joined
17 Aug 2005
Posts
2,825
Location
SW Scotland
If it was me. I'd download Afterburner. See what the default voltage displayed is (probably around 1v). Unlock the voltage control on the settings / general TAB, reboot Afterburner. And then try increasing it to something like 1.024v (completely safe at this, quite a few cards run at this voltage anyway). Remember to press "apply" button in Afterburner. Re-test and see what happens. Can't see you can loose anything.

I'm not guaranteeing that this is your problem. But there does appear to be a fair few graphics cards knocking around at the moment that don't seem to run 100% stable at the stock voltage! Just have a look at the Gigabyte, or MSI forums, lots of moans about 560/570/580 cards (especially OC cards)!!! The card I had problems with was a Gigabyte 580 OC (and that only had a tinnny OC on it).

PS. There was a problem a while back that for W7 DXdiag reported the wrong version for DX11 cards (reported DX 10.1 in fact). There was a hotfix ages ago to sort this.
IE. It was only showing the wrong version number.
 
Last edited:
Soldato
Joined
17 Aug 2005
Posts
2,825
Location
SW Scotland
Identical to mine, other than DDI version 11 on mine.

Might be an un-supported card for some reason!

Try Smart Doctor then. I would suggest 1.024v. Just double check that it has actually applied the voltage increase.

PS. Good luck. I'm off for a drink now. I'll have a look later to see how you got on.
 
Last edited:
Associate
OP
Joined
13 Jul 2011
Posts
16
Wow, I feel like I should give you the money I would have had to pay OcUK for the shipping and VAT to get it tested for faults lol.

I put it to 1.024v and it ran through the whole 3DMark 11 perfectly. I then went back to 0.975 to test again, and sure enough, it crashed with the error. I am once again testing at the new voltage but it looks like it will work!

I would never have thought of changing that because I would be afraid of killing the card :D.

But the ddi version is still bugging me. I CAN run the heaven 2.5 benchmark and enable tesselation, so dx11 MUST be enabled, I think.

Thanks again, you saved me from wasting money!
 
Soldato
Joined
17 Aug 2005
Posts
2,825
Location
SW Scotland
Good show.

I think DXdiag is just reporting the wrong version. I would tend to ignore it if it was me.

Right. You obviously have a decision to make now.....

Do you keep the card, or return it!

Obviously no problem for you to just run it at this voltage forever.
BUT... when / if you come to sell it on, you would need to mention that it wont run stable at stock volts and this might put someone off buying it. If this is a non issue for you, then fine. I was not willing to accept this on the card I bought, as it was mega expensive, so I returned it.

You could obviously edit the BIOS to make this voltage the default, but this is not the easiest of things to do.

In my opinion there are far too many cards coming through like this at the moment. MSI even offers a new BIOS for it's twin frozr 580 to "improve GPU stability" and all this does is bump the volts up a bit. Only my opinion folks, no flame war please.

Glad to help.
 
Last edited:
Associate
Joined
29 Apr 2014
Posts
1
Mikeo, i registered on these forums only to whorship and thank you. During months i've tried nearly everything with my 560 ti, nothing worked. Even getting a new power supply and reducing its cores. I was already researching for a new gpu when i decided to try what you said, and now i was finally able to run a full 3d mark test. Thanks you so very much!
 
Soldato
Joined
17 Aug 2005
Posts
2,825
Location
SW Scotland
Mikeo, i registered on these forums only to whorship and thank you. During months i've tried nearly everything with my 560 ti, nothing worked. Even getting a new power supply and reducing its cores. I was already researching for a new gpu when i decided to try what you said, and now i was finally able to run a full 3d mark test. Thanks you so very much!

No problems... Glad it helped you.

Though obviously being worshipped is no bad thing :)
 
Back
Top Bottom