So little while ago I made a thread about my EVGA GTX 680 SC not boosting as high anymore. 
The reason is that the card not only has a high temp throttling point at 70c and 80c, but has also a low temperature throttling point. At least on my card this point seems to be at around 43c. My card is water cooled and stays under that temp at all times, unless I take drastic measures to raise it above that. 
 
At 43c and under the card stops using the 1.175v and highest boost clocks, dropping to 1.165v and from 1175Mhz to 1150Mhz boost speed. Even when I manually adjust the voltage to 1.175v the card is not able to maintain quite as high OC as it did with the stock air cooler. My theory is that the throttling somehow still hampers the performance. Perhaps the real voltage is little bit higher when not throttled despite Precision X showing 1.175v in both cases.
 
In essence the performance of the card went down when I installed the water block on it. 
EVGA and nVidia are aware of this issue as I have a PM from Jacob stating so.
 
I find it odd that I haven't seen no one else reporting this issue. There should be other people with good water cooling solutions and a GTX 680. IMO this card is currently not very well suitable for water cooling if your cooling is powerful enough to keep the temps under the throttling point. Locked voltage control also takes the fun away from having a powerful cooling system, but at the very least I should be getting identical performance to the stock air cooler. I lose about 15-20Mhz in max OC currently, which is not too significant, but still annoying.
 
Jacob told me that they are working on this with nVidia and I'd hope to hear more about this and whether some type of fix is in the pipeline.

This is interesting and first I have heard of it. No wonder EVGA has not had the 680 Hydro Copper back in stock.
 
Wish I had hung onto my two vanilla 680 that clocked to 1255Mhz and put blocks on them instead of selling them to get the HC version.
 
Anyways, it sounds like to me these cards ignore actual graphics workload and base themselves off strictly the temperature. Even though your card may be at full stress load, the temp is telling the card otherwise so it down clocks itself thinking there is no need for the card to be clocked that high since the temp is so low it can't possibly be under full stress load.

kram36

This is interesting and first I have heard of it. No wonder EVGA has not had the 680 Hydro Copper back in stock.

Anyways, it sounds like to me these cards ignore actual graphics workload and base themselves off strictly the temperature. Even though your card may be at full stress load, the temp is telling the card otherwise so it down clocks itself thinking there is no need for the card to be clocked that high since the temp is so low it can't possibly be under full stress load.

Yep I also find it likely that the lack of HCs has something to do with this. It feels like a weird design oversight.

I have to say that perhaps the GK104 chip with it's mid range roots is partly the reason for these issues. Imo it was originally designed to occupy a different market segment, but with Tahiti offering less than expected performance and possibly issues with the GK100, we got this smaller chip as the high end chip, but it's old "shadow" still remains.

I'm still liking the card, but really wished it would stop throttling and that I could adjust voltages. I bought the card before I had any plans to build a water cooling into my system. Now I'm thinking I probably should have waited for the GK110 based cards and hold onto my 570 a year longer.

That's kind of messed up..
 
(gotta love this when using the evga website though:
Troubleshooting info:   Buffer overflow attempt detected. (201206290811231416102))
 

My GTX 680 FTW would do this to at around 63-64 degrees's. This was one of the issues it had that cause me to RMA it.

Try using very high target frame rate in Precision. Set it to around 100fps. I hope this helps.

rwoodburn1


Try using very high target frame rate in Precision. Set it to around 100fps. I hope this helps.

I had tried that with my card (except set it to 120 instead since I have a 120hz monitor) and it didn't help. I hope that can help the OP though.

rwoodburn1


Try using very high target frame rate in Precision. Set it to around 100fps. I hope this helps.

 
I've always used the unlimited uncapped framerates, so the card is not throttlling due to that setting. I did try it now though with no effect.

I realize the age of the thread but I 100% agree with you. I have a EVGA 680 sc sig on water. its in loop with my amd 1090t. normal temp is 38 and full load is 50. I cant OC more than +20 on the core otherwise it actually runs slower when using tests. It will never exceed 103% power usage and changing nothing simply closing OC scanner and pressing test again will always give different clocks. voltage is all over the show. I have tried everything from beta drivers, precision reinstalls, fan profiles (even though i dont have one) in desperate attemps to find my true overclock value.

Bump! I got the same issue, in another thread. Anyone heard of a fix, if Jacob is aware of this?

I thought this was fixed already but let me look into it some more...

Thanks so much, this is driving me mental :)
 
If you want I can send you the EVGA Precision hardware monitoring log, it shows the temperatures the card toggles clocks at..

Interesting; I have a GTX 680 water cooled and it runs at 40C under load and I've never had a problem with throttling or not being able to up the voltage either. Have you tried the Precision K-boost under the Voltage tab? Just a thought if anybody on WC with the 680 is having problems like this.

Can you use EVGA Precision voltage tuner and force 1.175V?

Hi,
 
I noticed I have not been really clear as to what I have done and at what stage I am of the troubleshooting, sorry about that!
 
Here is a recap:
 
Forcing voltage to max and with K-boost on in EVGA precision, card works IF IT IS OVER 26C!! (Important! - this is my issue ). It works both stock clocks and slight OC's, as it should be.
 
Even though I really press my system my idle for GPU will be 23-26 (even forcing the speeds) due to my somewhat.. overkill watercooling :)
 
As soon as I leave my PC or surf the web for a while the card clocks down/up according to temperature in the ranges of 23-26 ( 26+ makes it go stable at stock speeds )
 
I got hardware monitoring logs made with EVGA precision if anyone is interested. Sometimes the card state get's "locked" in 324mhz mode, and I have to uninstall drivers/drivercleaner/uninstall evga precision/reinstall drivers for it to work with K-boost and stuff again.
 
A bit of the log, edited; (OC'ed a bit)
Clock @
26C ; 1201.935
25C ; 888.387 ; 731.612 ; 614. 032 and finally stabilizing at 600.967
24C ; 574.838 then 535.645 ; 483.387 ; 431.129 ; 378.870 ; 339.677 and finally stabilizing at 324 

 

Ah, thanks for the clarification. That is very odd that even with the K-boost enabled you're having that problem. When I've set the K-boost on mine it stays at that voltage, but at idle I'm averaging 28C in a 27C office. I hope somebody has an answer for you because I'm just confused at this point! Good luck.

I did manage to run Battlefield 3 today for about 5 hours, but suddenly while browsing in windows the card locked itself again at 324 mhz, and nothing I do now can get it back. 
 
The card was working without K-boost, any OC, or alterations whatsoever. Why is this happening, I'm just getting issues all the time - seems like a faulty card.