OK so I've been over clocking graphics cards for years and I remember how it used to be back in the day, you clocked up high untill you got artifacts or the system locked up and then you brought it down a wee bit, easy. But now that I have just got myself a new MSI GTX 560 ti I am having a really annoying problem where I can't actually find the max overclock for my card as I keep getting the nvidia driver crash/recover mumbo jumbo whenever I start to run 3dmark or play some games. I can get upto about 950 on the core no problem without the driver crashing but any higher and it crashes before I've even had any artifact or system locks. So what is this driver crash crap the new way of finding out a max overclock or something and is their any way to disable it because it's really annoying me now. I'd rather just have the system lock up or get artifacts as that way I atleast know that it's overclock related and not something built into windows. Just to add my temps never go about 60 degrees c so temp isn't really a problem and I have good airflow in my case. Any help or input would be great, oh I'm on 275.33 and have not tried any other drivers. I only installed windows 2 days ago and had no previous drivers installed other than the default windows ones which I removed with drivercleaner.net. Thanks.
what program are you using to bench your card? what voltage do you have your card set at? what msi card do you have?
950 is really high, the driver crashes as a result of your overclock. Sometimes you get artifacts, more often just straight crashes. If it's crashing, you've gone too far.
I'm from the red camp myself (we have our own set of problems ) But a friend of mine had the exact same thing. In my search so far, everyone having this problem has an OC version > 900Mhz, I assume it's some kind of killswitch inside the driver. Why you run furmark / play games the Temp hangs @ 60c ? Thats seriously good airflow =] (my 2 5870's do 80/85) Heat isn't the problem, it seems to me it lacks enough power to handle the overclock. Try (just for the fun of it) to pull it back to 800'ish and torture it. See what happens. Due note: Back in the day artifacts made things pretty clear indeed, seen my share of them. Stop, adjust, retry. Until that one day... 5 days after my waranty expired... my Asus 8800 Ultra (Or Oeltra as guru called it winningly) died on me. Green-bars-of-death all over the screen... Still the only card worthy of a funeral if you asked me.
Its actually cuz Windows 7 its more robust, so instead of massive lockup, the driver crashes then Win 7 restores it. So if it crashes, it means that its unstable, with lock up/artifacts or not. 950Mhz is good. Most ppl reach around 1Ghz (that is, between 950Mhz and 1050Mhz) with some voltage tweaking and thats it. Dont expect 1,3Ghz or something...
My card is the MSI GTX 560 ti Twin Frozr II OC. I can get to 950 on the core with stock voltage and that's stable but really I'm aiming for 1GHz. If I put the voltage upto 1.1v I can get semi stable at 1GHz for a while but it will still crash sometimes. I am using 3DMark 11, 3DMark Vantage and Kombustor to bench my card. If you guys say that the driver crash is just Windows 7 way of telling me I've gone too far then that's my question answered, was just expecting hard locks and artifacts instead of driver crashes. I reckon I'll try for 980 on the core and just try and set my voltage accordingly, thanks for the replys and any more feedback is welcome.
I doubt a 50mhz bump would show any major improvement....but some just like to find the maximum clock rates.
I'm not having display driver crashes per say, but it resembles it. It seems to be related to the voltage and not the core clock frequency. When I play Crysis 2 at 1.1 V with the clock at 936 MHz it will rarely cause the core clock speed to drop to 405 MHz and voltage drops significantly. It's random, but rare. It hasn't happened yet in Crysis or Crysis Warhead, but it may just be luck. The game becomes unplayable at this point, and it won't adjust until I restart the computer. What could be causing this? Is the temperature of the voltage regulators getting too high? The GPU temperature is in the low 80s though, so I doubt that it could be related to high temperatures. My power supply is more than enough to support my overclocked system. Thinking it was the clock frequency, I lowered it from 942 to 936. However, it still occurs. Now I lowered the voltage to the next lowest step of 1.087 V. I'll see if it will still cause the odd drop in core clock frequency and voltage, but so far it has been stable. What causes this? ASUS GTX 560 Ti Forceware 270.61
your original voltage was too much for only being at 936mhz on the core. what type of 560ti do you have and what is your stock clocks?
@TheFool could be power limiter or not enough Amps on your psu.. im leaning towards your psu since Crysis 2 shouldn't be in power limiter list. For example; My oc 900mhz was 100% stable in 3dmark and heaven 2.5, vantage, JustCause2 but when i tried TessMark 4 i saw that gpu started to throttle.. Now either right away or after 40-50sec and raising voltage didn't help, but if i backed off to 887mhz and -16mv lower voltage it ran fine even after 120sec no throttling going on - min fps pinned to 74fps vs with throttling min 55-60fps and i could see that ball stuttering.. btw its best to use at least 90sec to see there is no throttling(i tested it at 2k maps, insane tess. 64 samples, with 8xmsaa and post fx2).. So that made me think its either power limiter with anything beyond 890mhz or my psu doesn't have enough amps or 12v rails.. weird because vantage and 3dmark2011, Heaven 2.5 passed without any stutter..guess this tessmark takes huge stress on both gpu and psu. And I got this driver stopped responding 1-2 times in Heaven 2.5 benchmark, i was testing 900mhz and it turned out i needed just a little more voltage. I raised it for another 6mv from 1.056v to 1.062v and that did the trick.
Here is my graphics card along with the specs and everything in the details tab: http://www.newegg.com/Product/Produ...&cm_mmc=TEMC-RMA-Approvel-_-Content-_-text-_- Here is my power supply, 630 W continuous or about 730 W peak (which is what most power supplies advertise as): http://www.newegg.com/Product/Produ...&cm_mmc=TEMC-RMA-Approvel-_-Content-_-text-_-
Personally, I wouldn't trust Rosewill (Newegg's "house" brand) .....I rank them up there with Dynex and Insignia (BestBuy's store brands)...
you have the same card as i do. anything past 960mhz is going to be a little touchy. if you look at my sig, ive got those clocks set at 1075mv. so that should give you an idea of where to take this card.