Official GTX 680 overclocking Thread!

Same for me for both cards. I think the default boost should be the same for everyone because that's what the cards are rated at.

woa now don't condemn the rest of us because you got cards that are less good. That is what is known as sour grapes I believe.
 
How does increasing voltage work? Does it actually increase to help with the core clocks or what?

According to Brent, it just raises the lowest voltage the card will drop down to. So instead of idling at 0.975, you can force it to idle at 1.075 or something - but the cards still max out at 1.175. He did say it can help with memory overclocking though, even if it doesn't raise the load voltage any.
 
According to Brent, it just raises the lowest voltage the card will drop down to. So instead of idling at 0.975, you can force it to idle at 1.075 or something - but the cards still max out at 1.175. He did say it can help with memory overclocking though, even if it doesn't raise the load voltage any.

I noticed that if I locked my voltage at 1.175 my boost goes WAY down. Anyone else getting this?
 
How do you detach the performance log like in the [H] review ?
 
woa now don't condemn the rest of us because you got cards that are less good. That is what is known as sour grapes I believe.

I didn't read thoroughly enough then, my apologies. I was referring to the default boost and not the OC potential.

I got great cards so no sour grapes here. I'm able to OC to +181 clock offset and +575 memory offset and it's stable through all benches. These settings leave the top card to go 1290 and the bottom to go 1320.
 
My high flow bracket came in, I can now stay under 70C in benchmarks. Getting 3-5C cooler in a lot of places.

Probably going on water soon anyway though or maybe when the next batch of 680s come out, maybe a 4gb version

Anyone have a preference on which block to use? Was going to go with the EVGA hydro copper and save myself the hassle of installing, but I do have a lot of IC Diamond lying around that I could put to good use.
 
I'm guessing that the 200 free brackets that EVGA was giving away are long gone by now?
 
Of course overclocking is never guaranteed, but on the other hand, it kind of sucks that we all put down $500, but some people seem to be getting cards that naturally boost higher, before overclocking. It just worries me a little that there's only 40-50 MHz headroom in my card -- in 12 months will it be crashing at stock speeds? I hope not.
 
Water cooled, 1310mhz running loaded. Runs about 37-38c.

overclock.jpg


IMG_0206%20%5b1280x768%5d.JPG
 
I think I got a dud. :(

Max OC I am able to achieve with fan at 75% (66% also works fine and is much quieter):
132% (haven't tried lowering it)
90 (cannot handle 100)
360 (might be able to get 390)
 
I think I got a dud. :(

Max OC I am able to achieve with fan at 75% (66% also works fine and is much quieter):
132% (haven't tried lowering it)
90 (cannot handle 100)

Haha, better than mine. Max fans +70, can't make 80. Wanna trade? :D I settled on +30 / +150 stock fan/voltage settings.
 
I think I got a dud. :(

Max OC I am able to achieve with fan at 75% (66% also works fine and is much quieter):
132% (haven't tried lowering it)
90 (cannot handle 100)
360 (might be able to get 390)

Have you guys tried re-applying the TIM? Maybe you've got a contact issue with the core?
 
Have you guys tried re-applying the TIM? Maybe you've got a contact issue with the core?

Wouldn't our temps be overly high if that was the case? Right now, running maxed Heaven, it gets to about 65*C with fan at 66% with the OC I posted previously.
 
Wouldn't our temps be overly high if that was the case? Right now, running maxed Heaven, it gets to about 65*C with fan at 66% with the OC I posted previously.

Maybe, sometimes contact with the core (whether CPU or GPU) stability could be affected before the temps are even an issue. Just wondering as I've see alot of well clocked 680s and some really bad ones. Just wondering if some have bad contact. Its a few screws and the stock tim is like toothpaste so I'd definitely recommend people check it and apply some decent TIM in any case.
 
Have you guys tried re-applying the TIM? Maybe you've got a contact issue with the core?

Full fan mine is at 57C when I hit the limit, and it crashes in seconds. It doesn't have much time to even warmup. Normal stock usage is 79C. I don't think TIM is going to help here.
 
+150 +450 here with an actual core clock of 1240 in BF3. This is with no manual voltage changes, only the 132%TDP slider.
 
Quagmire LXIX- lmao.

MavericK96 - no ram sinks, 120 mm fan blowing over the card though.
 
Actually, shouldn't we be setting power to +70%, and then adjust the +clock? That'll peg it to a more narrow range of frequencies, right? That would basically disable boost, which doesn't matter because you're manually increasing the frequency anyways. Which cures the "oh no must have 70C for max boost" thing. Or have I missed something.
 
Here is mine...I'm very happy with this OC. these settings give me 1245MHz in all games.

Power Target 125%
GPU Offset +121
Mem Offset +480
 
Time for me to try for +95 and +385. :cool:

Edit: Seems to be stable.
 
Last edited:
Actually, shouldn't we be setting power to +70%, and then adjust the +clock? That'll peg it to a more narrow range of frequencies, right? That would basically disable boost, which doesn't matter because you're manually increasing the frequency anyways. Which cures the "oh no must have 70C for max boost" thing. Or have I missed something.

I don't think that would work. The card increases frequency as much as it can and remain within the power limit you set, so setting 70% power limit, instead of 132% (if that is what you are saying) is just going to limit the ability of the card to boost. It wouldn't matter what +freq you added, because the card still won't let you exceed the power target you set. With my card setting +150 pushes my card to about 120% power limit in BF3 - if I reduced the power target to 70% I believe the card will just chop the base freq as much as required to get that power target to 70% - so instead of running at 1110+150, it would run at 700+150, for example.

Think of it like Sandy Bridge overclocking, you aren't setting an actual frequency for the card to work at, you are just changing the boost paramaters (like changing the turbo multipliers in SB).

Might give it a whirl when I get home though, just to see what happens.
 
It wouldn't matter what +freq you added, because the card still won't let you exceed the power target you set.

That's not what I saw, but I can't test anything over +50 with my card, so you'll have to try it out.
 
That's not what I saw, but I can't test anything over +50 with my card, so you'll have to try it out.

So if you set 70% power target, it was running in excess of that?

Edit: Tried running at 75% power (it wouldn't let me set 70 for some reason) and it limited my card to the normal boost speed even if I had +200 set. It also ran at 100% card power - so it appears that setting a power limit below 100% doesn't affect factory performance, but just eliminates overclocking. Maybe new drivers/versions of PRecision will change that.
 
Last edited:
I was using MSI AB. Didn't like PrecisionX. (I believe it's the same guy behind both, but they are slightly different in how one sets the settings.) Yeah, 100% gpu power at 70% setting here.
 
I was using MSI AB. Didn't like PrecisionX. (I believe it's the same guy behind both, but they are slightly different in how one sets the settings.) Yeah, 100% gpu power at 70% setting here.

I had trouble getting Afterburner to hold a power setting below 100%, so I tried PrecisionX. Normally I use Afterburner as well.
 
Full fan mine is at 57C when I hit the limit, and it crashes in seconds. It doesn't have much time to even warmup. Normal stock usage is 79C. I don't think TIM is going to help here.

My card rarely goes over 70, highest I've ever seen is 72, I think? So maybe your TIM is, indeed, poopy or poorly applied? 79 seems a little warm comparatively.

Granted we could have different case cooling or ambient temps.

Just a thought!
 
I managed to pick up 2 EVGA GTX 680 SC's last week and have had very little luck in overclocking. I only went with the SC cards because they were available at the time and were the first of four GTX 680 orders I placed that didn't get voided, cancelled, or back-ordered.

Since these cards already effectively start off with +66 on the GPU and +100 on Memory I decided to start small, +50 GPU, +100 Memory. This always resulted in a crash or the screen going black, the monitor losing signal for 2 seconds, then the GPU's powering back up with stock clocks again. So far, anything over +20 on the GPU is eventually unstable. That's effectively +86 on the GPU's in SLI.

Memory can go much higher on its own depending upon the game. In Skyrim I can go up to +400 before I get artifacts which puts me at +500 effectively. This usually causes a crash though when combined with the GPU offset so I have to dial it back to about +300.

Several games play fine at +20 / +300, Stalker, Metro 2033, and Skyrim for example. Path of Exile on the other hand didn't like any overclocks and proceeded to crash or perform the reset cycle with anything over stock. So... it's hit or miss.

Temps are great so far. I have the fans locked at 70% since, at that point, they are as loud any other fan in my case and I haven't seen the cards go over 60C. I've also recorded the total system wattage and the numbers are ridiculous. In Metro 2033 with both GPU's overclocked and running at 95% usage or more my entire PC was only drawing ~ 475 watts from the wall!. That's less than my system used to pull with a single stock GTX 480.

Thankfully every single game I've tested has played beautifully in SLI at stock settings. It seriously feels like a major step up from the experience I had with crossfired 6950's.
 
Last edited:
Back
Top