When I apply my regular oc settings the forceware driver imminently crash - latest afterbruner version
and please install 314.09 WHGL, they have fixed many problems with 3D Vision/Surround (even on Sli) ... yes it's the driver for Titan but works great on other cards to
what are u talking about? "find a stable OC" the OC was extremely stable over a year and its not even very high - its not working since this new driver
Different drivers are different. You have a choice, be moody and have no fun. Or find the best overclock and have some fun. Theres also the possibility that a little aging has set in, reducing stability just at the time you changed driver. Or that your room temps are a little higher than normal. Or your PSU isnt as stable as it used to be. In which case you still need to find a new stable overclock. Or your fans/heatsinks have a little dust build up and need a clean. (but this should be checked before making a thread about stability) ps another thing to try is raising the voltage slightly. You may get the same effect by speeding the fan up.
Mostly its this. For me 304.xx and later needed extra 6mv to be stable at 900mhz, its been fine with any driver after that.. Before 304.xx i was also 1000% stable with 6mv less. I don't run my OC 24/7 only for certain games, I leave it at stock clocks/voltage when in idle or less demanding games. Also vram OC varies a bit, latest 313-314.xx can keep up with higher OC, I couldn't do that with any 310.xx or few 304.xx.
for me it seems 314 can't handle as high OC or the driver just has issue in FC3 . I have been running mild OC 1228mhz and right after 314 I started getting artifacts, texture corruption in game randomly , restart would fix and be fine . Anyway seems lowering OC back down to 1202mhz (stock OC ) seems to be more stable , need more testing to be sure .
Far Cry 3 made my GTX580 the hottest I have ever seen it. Most games are 55 to 65C when overclocked, FC3 took it to mid 80's ! I had to reduce my overclock to 870/4600 1.050V on the core. I've seen 72C with Crysis 3, this game is stable though at 920/4600 1.125V. I previously tested my card to be stable at 930/4650.1.125V, so thats probably why I havent had to make an adjustment for this game, or for different drivers. If you push the ragged edge, different drivers definitely give different max clocks.
I have MSI 660ti PE/OC and it has good fan setup , all I needed to do was setup fan profile in AB so 50% fan gets used instead of 40% before . I never even hit 60c in FC3 now, its normally 55-57c . This game is weird though even in menu screens the temps are high . Oh, I found to to get max clocks it seems you need temp between 60-70c , I don't care about it, I rather run cooler and just off-set the clocks higher in AB . Something for sure has changed IMO with latest drivers . In FC3 I was not running max clocks before , I could run 1254 if I wanted but don't see reason to push it like that . Anyway looks like you just have to tweak them after newer driver series .
Driver devs have to balance heat with the amount of processing. There are numerous trade offs they can employ, including using the CPU a little more to alleviate the load on the gfx card a little, for processes that lend themselves to this. Sometimes its possible to get higher performance from a card by removing redundant steps which can result in higher performance and lower heat. But they dont come along very often as they take a long time to find and get stable. This can be offset against other performance increases that add more heat, so it can be a double win when these are found. Much of the performance increases they can provide will cause an increase in heat on some part of the GPU. They may be able to trade off processing on one part of the GPU for processing on another and its not until this goes out into the wild that they know which is the preferable solution. An increase in heat production may not be so bad if it is produced on a part of the GPU that is running below spec, but it can still affect other parts of the GPU as the heat spreads. Then theres the question of whether the voltage is high enough, can the vrms supply enough current stably and will the vrms get too hot. Will a new fan profile be needed? This is one of the reasons for beta drivers, so things can be tested. There are so many different versions of each card with different quality ASICs, different power circuits, different cooling solutions, used in different environments that it literally is often suck and see. So for sure beta drivers can have a wide variance in temps and hence stability with temps. WHQL are generally made from the more stable beta drivers. But then theres the competition element. When ATI find a way of improving performance through their driver and release it, NVidia respond by releasing a driver with a performance increase they have kept for such a reason , and vise versa. On occasion, to compete, the WHQL drivers have caused an increase in heat that require us to re-evaluate our overclocks. They also make some pre-overclocked cards need an RMA or a downclock if out of warranty. Then theres games. If a certain set of instructions that create more heat is used more by a new game, NVidia may have to create a driver to prevent problems.
For sure there are a lot of variables both in the driver to app/game and then on your system . I don't know if it is true, on another forum some were saying the voltage has been limited with 313+ , I don't know even if that was true if it would affect me with stock bios (1.175v) , I could see Nvidia limiting higher in some way but the whole power-management is very hard to test on our side .
