EDIT: RMA'ed:Galaxy 660TI - artifacts - randomly happens - HELP!!! What to do next?

Engineer

Elite Member
Oct 9, 1999
39,230
701
126
I've had this card for a few months but not really ran until the last 3 to 4 weeks. It's a Galaxy 2GB 660Ti card. It runs fine 99% of the time but in BF3, it will all of a sudden throw up artifacts. Once it does, they also appear in other benchmarks such as 3DMark11. A simple "warm reboot" (START, Restart) corrects it.

Card: http://www.newegg.com/Product/Produc...82E16814162120

My Power supply is a new Seasonic X650 GOLD. I have 32GB of Patriot DDR3 1333 that just had 2 sticks replaced because they were bad (now functioning fine under memtest86+). Board is an MSI Z77MA-G45 (flashed with latest BIOS). The CPU is a XEON E3-1230V2 running at stock with a Xigmatek 120MM Dark Knight cooler.

Everything is running at stock.

Ran FurMark for nearly 2 hours and temperatures of the GPU never went above 74C and no artifacting. Ran VMT (Video Memory Test) and it found ZERO errors in ram (have not ran either once the artifacting starts - hard to know when that will happen). OCCT GPU memory test reports thousands of errors but I'm not sure that which test to believe - VMT or OCCT since one says good and the other says bad.

Have swapped the card to different slots on the motherboard and have tried 2 generations of drivers as well as the latest BETA (actually, have had no artifacts during the BETA but it doesn't happen often so no proof just yet).

Anything else I can do to test this? I don't have a power supply right now but I could, if necessary, swap it out to another PC with the same supply in it. I'm just at a loss as to what's causing it.

While my main memory was taking a dump, the video card did stop responding several times with Windows posting that the video card driver had stopped and was restarted. It has not happened since I replaced the ram and reinstalled the driver. I checked the Windows system files with SFC (System File Checker) to make sure that no files had become corrupt due to the bad memory. Integrity was reported as 100%.

Any other ideas short of pulling the card and RMA just to be safe?
 
Last edited:

Anarchist420

Diamond Member
Feb 13, 2010
8,645
0
76
www.facebook.com
check the official geforce forums to see if anyone has had similar issues to your's. I'd guess that it's the driver, I have an eVGA 660 Ti with the stock pcb and cooler and I haven't had any of those problems. I'm using the latest beta driver (i did a clean install) although I've never played battlefield 3.
 

Engineer

Elite Member
Oct 9, 1999
39,230
701
126
I read around over there and never found an answer. With that said, either the BETA driver fixed it or something else did OR it's just hiding and waiting to come back out.

I've run the FurMark burn in test for over an hour (80C this time) and still cannot get artifacts in BF3 or anything else. I have done nothing other than install the BETA driver (now WHQL version 314.22 - already upgraded).

I guess I'll monitor this and see where it goes.
 

FalseChristian

Diamond Member
Jan 7, 2002
3,322
0
71
Do NOT use Furmark! Do you want to fry your GPU? Besides, NVidia's drivers throttles the card or else it would overheat to dangerous levels.
 

Engineer

Elite Member
Oct 9, 1999
39,230
701
126
Do NOT use Furmark! Do you want to fry your GPU? Besides, NVidia's drivers throttles the card or else it would overheat to dangerous levels.

OK, I'll keep that in mind. With that said, the card never went above 80C and the fans to 53%.

The problem started before the FurMark tests though and I just found it: BAD MEMORY that goes in and out. I left the DOS based memory test (not a stress test, just a pattern test) open in a command prompt and was randomly running it over the last few days. I did it just now and had about 10,000 errors in the first GB of ram on the card. I immediately ran 3dMark11 and BF3 and both were loaded with artifacts.

I have already contacted Galaxy for a RMA (although it could be my new Seasonic Power supply). I might swap it to another PC for a few days just to make sure before sending back (*sigh* - I hate to send stuff out of my PC and have to tear it down and build back up several times *sigh*).

Anyone have any thoughts that it could be the power supply causing issues? I did have 2 of my sticks of ram go bad in same system (which is around 2 months old now)?

How to test power supply?

Also, why would a "warm" reboot fix these errors temporarily?
 
Last edited:

Engineer

Elite Member
Oct 9, 1999
39,230
701
126
After doing much testing, this appears to be "Sleep" related. The only time that the memory starts having errors is after the PC wakes from sleep. I can reboot (warm or cold) or even put the system back to sleep again and it will work correctly (no mem errors) 99% of the time.

Not sure if this is a BIOS issue, Windows issue or driver issue. I've done everything Galaxy has asked other than to swap power supplies and I'll try that this weekend.

I don't think it's the supply.
 

Engineer

Elite Member
Oct 9, 1999
39,230
701
126
Swapped power supplies last night and, initially, all was well. This morning, woke up the PC and had memory errors "until"....the driver crashed and recovered and then no more memory errors. This is starting to sound like a software issue more than a hardware issue as a simple crash and recover would not recover hardware, no?

Thoughts? I'm on the verge of RMA'ing it just in case but if I get another and it's still doing it, not sure what to do next (other than FULL reinstall).
 

Engineer

Elite Member
Oct 9, 1999
39,230
701
126
Question: If you take a screenshot of something that artifacts, should the artifacts show in the screen shot? The reason that I ask is that I read that if you take a screenshot and the artifacts show, then it is the drivers fault. If the artifacts do not show, it is the cards fault (bad mem)?

I took a screen shot this morning and the artifacts show. Would this be a driver problem or a memory problem? Can a screen shot (<ALT><PRT Screen>) capture memory errors on a video card?
 

Engineer

Elite Member
Oct 9, 1999
39,230
701
126
Update: New card in and have not been able to duplicate the issues/memory errors so I'm assuming that it was indeed hardware. Will try to duplicate (not stress test, simply monitor) the errors to make sure that it wasn't my system (OS, RAM, board, etc).