RMA - No fault found - advice?

avrc · 12 Jan 2011 at 20:42

OLDPHART said:
Was that one stick at a time tested, if it was put them both in and run memtest to see if they pass.

That was both together. It's passing fine on 1333. I have read a review elsewhere where someone couldn't get it running stably at 1600 and had to use 1333.

I guess I'll see what happens when I get it back.

Garido · 12 Jan 2011 at 21:00

In any case the RAM is now confirmed as safe and passes memtest which rules that out. Concentrating back onto the card, you have tried different drivers which hasn't resolved the issue or crashes. Ask the store what tests they performed and how long they were performed for. It's not like your card is DoA, rather it seems very finicky.

I wish you the best of luck. Keep us updated.

avrc · 12 Jan 2011 at 21:09

Not sure they provide a detailed test sheet but I shall ask.

This is what I was emailed which is a standard cut and paste job.

Here is the technicians Test Report:

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

"Graphics card passes all Tests with no issues, Tests Run: Crysis WarHead, Furmark, 3Dmark Vantage, AVP, Lost Planet 2, Stone Benchmark MSI Kombuster, Resident Evil Benchmark, ATI Mecha, ATI Ladybug, "

(Graphics card passes all Tests with no issues, Tests Run: Crysis WarHead, Furmark, 3Dmark Vantage, AVP, Lost Planet 2, Stone Benchmark MSI Kombuster, Resident Evil Benchmark, ATI Mecha, ATI Ladybug, Streetfighter Benchmark and heaven benchmark tool, these were all tested at 1920 x 1200 resolution running maximum settings, tested on DX9, DX10 and DX11 where applicable, there was no issues apparent.)

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

I would have liked to have seen times run for and temperatures.

Garido · 12 Jan 2011 at 21:30

When I worked in a PC store (not OCUK, not even in the UK for that matter) we ran Furmark as the test along with a few others. If the card was an NFF (no fault found) screens and photos were taken of the card in a test rig as well as the screen and timestamp as proof. In some cases a labour/service charge was taken by the store.

Perhaps you can ask for proof and more details on their tests?

Treznor · 12 Jan 2011 at 21:56

PhillyDee said:
Treznor - What exactly is the circumstances though?

crysis crashes when it loads into the game. The unigine benchmarks crash at the same point everytime. The windows experience index wont complete, the pc crashes on the dx10 geometry test everytime

There are other instances too, the weird thing is, it seems that the majority of games work fine, it just seems certain ones always crash 100% in the exact same place, so it is certainly hard to pin point where the fault lies, but if there was no fault then I would not be able to replicate the issue in a totally different system.

When I say crash, the application is not crashing, its locking up the PC, nothing responds, no mouse / keyboard nothing, only option is to power down and restart, I appreciate its not a straight forward issue.

Treznor · 12 Jan 2011 at 21:57

Garido said:
When I worked in a PC store (not OCUK, not even in the UK for that matter) we ran Furmark as the test along with a few others. If the card was an NFF (no fault found) screens and photos were taken of the card in a test rig as well as the screen and timestamp as proof. In some cases a labour/service charge was taken by the store.

Perhaps you can ask for proof and more details on their tests?

I can run furmark with no issues, yet I still have other issues with the card, so thats still not saying the card will work correctly.

Garido · 12 Jan 2011 at 22:04

Treznor said:
I can run furmark with no issues, yet I still have other issues with the card, so thats still not saying the card will work correctly.

If you read my post I said "along with a few others" and by that I mean we also ran a few other tests. If the customer specified a specific problem, we would try and replicate that.

My own 3870 in my AMD rig had an issue in which it would break up/stutter and lose sync. It reported at being 8x speed despite being in a 16x slot. Card was faulty but passed 3DMark06 loops.

In the end I asked the store, just play a game for 10 minutes. Within the first 5 minutes the card crashed, and I got my money back and went elsewhere (again, not OCUK was not the store involved, I live abroad).

Dervious · 12 Jan 2011 at 23:09

avrc said:
Hi Guys,

I RMA'd my Sparkle 460 GTX 1GB this week after a week of problems with it - every game I tried crashed with graphical errors either to the desktop or froze the system anywhere from 5 mins to an hour of game play. I found running the fan on max prevented the crashes but still produced graphical errors now and again.

I tried the usual - various driver and a fresh windows install. I swapped it out for my old 260 GTX and that runs with no problems.

Overclockers have emailed me to say they can't find a fault so I now have to pay £10, shipping and VAT on top to get it back.

Now, the card was an xmas gift so I'm not out of pocket at this point. But, I can't see the point in paying for the card back when it doesn't work for me.

I also emailed Sparkle but all they can say is that they've never heard of this fault before. That was it!

So, do I just right this off and leave it with Overclockers as it hasn't cost me anything or is there something else I can do? The card cost my father £141.99.

I nearly sent my Gigabyte GTX460 back to OCUK when I discovered it kept locking up after a few minutes of gaming or benchmarking. I then found that the card worked perfectly in 2 other PCs. If I had sent the card back to OCUK, I bet it would have been "NFF, £10 + shipping please". Long story short..... the GTX460 was faulty, and Gigabyte themselves replaced it. The replacement isn't perfect, but at least it's working in my current system.

Oh, and this may be of interest - http://forums.nvidia.com/index.php?showtopic=187401&st=0

avrc · 13 Jan 2011 at 00:28

Yeah I've seen that thread. I was going to post on the nvidia forum but for some reason I can't.

Unfortunately I have a Sparkle card and support seems non existent and comes via Taiwan so RMA'ing to them doesn't seem to be an option.

Not sure what I'll do with it if I get it back and it still won't work properly.

Rroff · 13 Jan 2011 at 01:09

Not saying it is or isn't the card thats faulty, but faulty RAM can cause graphical errors as it will be uploading faulty data to the GPU which causes the GPU to handle the data incorrectly. If its only a small number of faulty locations the difference between 2 GPUs could be enough to mask the RAM error due to the slight differences between the way the 2 GPUs work.

avrc · 13 Jan 2011 at 01:24

After reading through the nvidia thread it seems there is no solution. Whilst the thread starts out with extreme symptoms there are people posting with the same issues as I have. As it stands it seems to be a compatibility issue between cards and motherboards.

J.D · 13 Jan 2011 at 02:14

avrc said:
After reading through the nvidia thread it seems there is no solution. Whilst the thread starts out with extreme symptoms there are people posting with the same issues as I have. As it stands it seems to be a compatibility issue between cards and motherboards.

Well I've seen this a few times now and I'd hope if that card is less than 2 years old that OCUK sort this out for you. If I know about it through OCUK's forum then I hope they also realise this is a problem too.

Is the motherboard from OCUK also?. If so and the problem persists, send it all into them.

avrc · 13 Jan 2011 at 08:29

Mobo was from elsewhere. I sent OCUK a message last night linking them to that nvidia thread.

Reading that thread reminded me of a problem I'd completely forgotten and one that confused me at the time. With each driver I installed, (one on CD, latest stable and latest beta) the screen would go blank and then go into standby and not recover until the pc was restarted. I just thought it a strange quirk at the time but I've read other posts where screens have gone into standby.

Dervious · 13 Jan 2011 at 09:02

Rroff said:
Not saying it is or isn't the card thats faulty, but faulty RAM can cause graphical errors as it will be uploading faulty data to the GPU which causes the GPU to handle the data incorrectly. If its only a small number of faulty locations the difference between 2 GPUs could be enough to mask the RAM error due to the slight differences between the way the 2 GPUs work.

Agreed. Just because I, and many others have had problems with our cards, doen't mean that the OP's problem is the card. I had to spend hours borrowing/swapping components until I conclusively proved that the card was at fault. Even after changing everything (apart from my motherboard), and reinstalling Windows, the original card wouldn't work, but did in 2 other PCs. That made me certain it was my motherboard, which Gigabyte tested and returned to me. I then had to send both the graphics card and the motherboard, which proved that the graphics card was at fault. Now I have a replacement card that works with my motherboard, but it doesn't work in the other motherboard I own ! Luckily (?) for me, my 2 motherboards and my GTX460 are all Gigabyte products, so Gigabyte couldn't blame any other manufacturer.

It's been a bit of a nightmare, and has lowered my opinion of Nvidia.

mrfil13 · 13 Jan 2011 at 13:47

Graphics card issues as a real pain.

Running Furmark is not a clear way to test a card, i did it with mine for about an hour and it was fine, ran crysis for about 2 mins and it hung every time.

Ezyryder · 13 Jan 2011 at 14:27

Thanks goodness I never had that problem when I had my 460 GTX, sounds like hell.

Imagine what would happen if this was happening with an AMD card, jeezuz!!!

Mr Paul · 13 Jan 2011 at 14:35

Garido · 13 Jan 2011 at 15:51

mrfil13 said:
Graphics card issues as a real pain.

Running Furmark is not a clear way to test a card, i did it with mine for about an hour and it was fine, ran crysis for about 2 mins and it hung every time.

This is why you run multiple tests and not just Furmark. In my experience, most failures were detected quicker by running Furmark (artifacts and crashes) as opposed to a 3DMark loop or even running a game.

Not every fault will be the same

Redmint · 21 Jan 2011 at 17:07

avrc said:
Mobo was from elsewhere. I sent OCUK a message last night linking them to that nvidia thread.

Reading that thread reminded me of a problem I'd completely forgotten and one that confused me at the time. With each driver I installed, (one on CD, latest stable and latest beta) the screen would go blank and then go into standby and not recover until the pc was restarted. I just thought it a strange quirk at the time but I've read other posts where screens have gone into standby.

Just wondering what happened with this?

avrc · 29 Dec 2013 at 13:42

Redmint said:
Just wondering what happened with this?

Old thread I know but I started it and finally have an update after all this time! I ended up buying a Radeon card in the end to replace it and the 460 has been sat in a box since then.

My step son bought COD Ghosts for Xmas but didn't read the specs for it so I've had to fiddle with his machine - swap to 64bit windows, add more memory and then realised his old graphics card didn't support DX11 so I thought I'd dig out the 460 and give it another go. I had previously tried it with this machine and had no luck. To cut the story short it seems to be a contact issue between card and slot (I'd tried this card on 3 different motherboards before with no luck) anyway I realised it worked when the machine was on its side ie motherboard horizontal (which is probably how they get tested at ocuk?) but as soon as the machine was stood up it filled the screen with artifacts. I used a bit of garden twine and supported the back of the card via a drive bay and hey presto no more artifacts. Latest nvidia driver installed and he's been playing the game none stop for a few hours now without issue.