i've oced my gtx560 from 850 core clock and 2050 mem clock to 931 / 2198 with 1.050v... but an hour ago after logging in to WoW terrible artifacting suddenly had befallen me. i ran evga OC scanner and with my OC it shown terribly artifacts and the same with furmark. i've uninstalled the driver with driver sweeper and installed it again (296.10, 301.42 is giving me much more problems). still it persists. even if i lower the clock to 900 / 2152 / stock volt, the problem still persists (extreme artifacting leading to drivers crashing/BSOD). but whenever i put it to stock(no OC or w/ever) it just run fine, no artifact or any crashing. what seems to be the problem? is it time for an RMA? this gpu is quite young, bought it last december 2011. i really really need your help right now guys!!!!

My first thoughts order:

1. Maybe it wasn't stable originally and you haven't stressed it fully yet
2. Maybe you discovered you don't have the best batch or ruined a good one
3. Maybe the summer weather higher amibient temps are affecting your stability and the GPU is relatively too hot.

• If #1, if you redo the same stress test you did before, you may discover the GPU needs a slight bump in voltage to keep it stable, which will raise temps further.
• If #2, you're out of luck.  I had a 7900GT KO once and had a higher air overclock than the average did with water, but it lasted only 3 months before I had to RMA due to artifacts, while the replacement wouldn't reach the water average overclock.  Humbled and lesson learned: I never expect absolute max for long-term stability, maybe I was asking for it.
• If #3, maybe you'll get your OC back by fall or winter, maybe changing your cooling will gain temps as they were, or turn on an AC unit in your room, or open a side door, or improve chassis airflow to lower the temperature of air entering your cooler's fan-- that is the local ambient temperature that may be warm.  Find a way to bring that down.

It was probably never really a stable oc and needs more voltage. Example with mine. I need 1.075v to hit 900mhz core and 2079mhz memory. My card tops out at 910mhz 1.087v on the core alone. I really seem to think its a voltage issue. Since it runs fine stock. Another possibility. As the drivers progress. The cards get pushed harder. So that can effect oc as well. Might only be stable at SC speeds. Also. If you RMA a perfectly working card and it works fine stock. Evga most likely would send it right back to you. OC is not a guarantee.

thanks for the reply guys! i am currently testing it under 900 core clock and 2152 mem clock with 1.012 voltage.. running OC scanner for 15 mins still no artifacts. i hope this'll went well.

ran OC scanner X for 1 hour and everything is fine with 900 core and 2152 mem clock @ 1.012volts. in the past i can get this on just on stock voltage. what could be the reason why?

OC scanner is useless accept for when a card is actually failing. I can run my oc on stock voltage and it passes fine as well. Mine is 1.012v btw for stock. Once I get in a game though. Especially BF3. Crash and burn. Same with heaven benchmark 3.0

One other thing to consider is, if your GPU runs fine at stock clocks, you have no valid issue for an RMA.  Secondarily, artifacting is frequently a memory issue. Overclocking your GPU memory adds little to performance compared to overclocking the core.

HeavyHemi


One other thing to consider is, if your GPU runs fine at stock clocks, you have no valid issue for an RMA.  Secondarily, artifacting is frequently a memory issue. Overclocking your GPU memory adds little to performance compared to overclocking the core.

 
so should i lower the mem clock? yes it runs quite fine on stock

FattysGoneWild


OC scanner is useless accept for when a card is actually failing. I can run my oc on stock voltage and it passes fine as well. Mine is 1.012v btw for stock. Once I get in a game though. Especially BF3. Crash and burn. Same with heaven benchmark 3.0

thanks for that... how bout furmark?

Furmark is no good either. Since Furmark draws crazy amounts of power. Nvidia set a limit in the drivers for it does not damage the card.

how can i test this?

raident30


how can i test this?

Essentially the only way to test for stability in games, is gaming. Furmark, OC Scanner, etc etc, don't stress the GPU the same way games do. BF3 and Crysis 2 are a couple of DX11 titles that are sensitive to GPU stability. The F@H client is also pretty good for checking over clock stability.
As far as your memory clocks, I've benched with my cores as high as 975 and memory at 2200. There really is no advantage going over 2100 or even a bit lower. For daily use I have them flashed to 850/2100 at 1.063 as I value absolute stability over a marginal increase in performance. I estimate my combined performance gain by overclocking the CPU and GPU is ~20%. Not bad for free.

Yea I've gone through a good amount of 560's and I can tell you right now; the card is unpredictable overclocking wise.

My first 560 SC was stable at 950/1900/2052 and I had to RMA because the plastic shroud fell off.. My second 560 SC was a dream and would overclock to 960-980MHz on the core @ 1.87mV till about 2 months or so later when it started crapping out, my third 560 SC was a DOA RMA, and my 4th and current 560 SC is giving me no room to OC with only 900/1800/2052 @ 1.25mV - 1.63mV just to get stable. And I still can't play BF3 without getting a BSOD due to memory problems on the GPU after about 3-4 hours later.

SAYSteez


Yea I've gone through a good amount of 560's and I can tell you right now; the card is unpredictable overclocking wise.

My first 560 SC was stable at 950/1900/2052 and I had to RMA because the plastic shroud fell off.. My second 560 SC was a dream and would overclock to 960-980MHz on the core @ 1.87mV till about 2 months or so later when it started crapping out, my third 560 SC was a DOA RMA, and my 4th and current 560 SC is giving me no room to OC with only 900/1800/2052 @ 1.25mV - 1.63mV just to get stable. And I still can't play BF3 without getting a BSOD due to memory problems on the GPU after about 3-4 hours later.

 
how long did that dream OCer 560 go? by the way you said crapping, what do you mean by this? and about your latest 560, do you always get BSOD from it? my 560 is 5months old now and i haven't got any BSOD until now. 

HeavyHemi

raident30


how can i test this?

Essentially the only way to test for stability in games, is gaming. Furmark, OC Scanner, etc etc, don't stress the GPU the same way games do. BF3 and Crysis 2 are a couple of DX11 titles that are sensitive to GPU stability. The F@H client is also pretty good for checking over clock stability.
As far as your memory clocks, I've benched with my cores as high as 975 and memory at 2200. There really is no advantage going over 2100 or even a bit lower. For daily use I have them flashed to 850/2100 at 1.063 as I value absolute stability over a marginal increase in performance. I estimate my combined performance gain by overclocking the CPU and GPU is ~20%. Not bad for free.

 
 it's quite stable at 900core and 2152 mem clock @ 1.012v, played WoW for 6 hours now and experienced no problem. gotta play crysis 2 later hope it'll be fine(quite dejected with bf3 cuz i haven't bought it's premium yet :((( ).. your talking about your 580 right? what is the stock clock of that? should i increase more voltage from 1.012 to gain absolute stability? did you encounter any problems with your 580?
 
 
btw, thanks for your help and support guys!

agreed, bf3 is really the best gpu-testing game of recent......... Metro 2033 works well too.

how about crysis 2 and max payne3? i don't wanna play BF3 these days( it became a pseudo COD)...