Im interested in TDP, fan doesn't bother me, great thread.

If you install Afterburner and change no setting at all except raise power limit from 100% to 110%, the card is a lot better at holding a flat 99% GPU utilization and you coincidentally get loss chop in game due to this.  The ridiculous part is that even with stock clocks on this card, the power TDP limit hit 111-113% in Lichdom, so there is not only 0 room to overclock on this card, setting it to maximum TDP power limit of 110% is still not good enough for stock clocks.  The other 970's I've tried are also able to hold a flat 1342 speed in games without fluctuating, while this card pings up and down still from 1390 to 1405 even with TDP slider at 110%.  The constantly pinging clock speed and anemic TDP value where you can't hit max GPU utilization makes this card stutter more than other cards.  The FTW needs a BIOS update to 120% TDP to be a valid card while using the stock clocks it ships with.  For people to have any overclock room at all, it would need 140%.
 

 
Here's what GPU utilization looks like with power slider at default, constant pinging up and down while sitting at 98% most the time:
 

 
 
 
 

Maybe the 970 you bought has to high of a voltage ID to run those clocks with out hitting the tdp i would send it back.
But then again I guess waiting for an answer would be best for sure.

freestylingford
Maybe the 970 you bought has to high of a voltage ID to run those clocks with out hitting the tdp i would send it back.
But then again I guess waiting for an answer would be best for sure.

It's a pretty normal VID chip...I'm just testing with a graphically intensive game.

I actually take back the comment about not being a complete stutter fest.  If you go to any wide open area that really taxes the card and not inside a cave, instead of being more stuttery but bearable than other 970s, the 970 FTW has crazy amounts of stutter due to the TDP issue.  I just need to know if they plan to release a BIOS fix for this or if I should return the card.

Isn't there a bios at overclock.net that sets the TDP to 250W?

r0ach
Just found out my FTW BIOS is limited to 187w with power slider at 110%, the same as the ghetto SC card...  That means this card has the lowest TDP out of all of them while having the highest boost clock, no wonder it has issues.  The ASUS Strix is the second lowest TDP with 196W at max power slider, but it doesn't boost anywhere near 1405.

 
Another user stated it has dual bios, have you tested both BIOS?  usually one has higher TDP and more aggressive fan profiles to allow for higher overclocking while being smooth.  
 
*heck* it has triple bios support, have you tried any of them?
 

rshwayder
It is. There's a small switch on the back.

Ugh, that dual BIOS post describing what each BIOS does is sure hidden well...Does the 2nd BIOS have a higher TDP limit than 187w with 110% power slider?  Guess I'll go look and see.

You said you overclocked a 970 FTW over 1500mhz...
 
How is this possible when the card has a 187w TDP limit and the MSI card with a 220w power limit throttles at 1500mhz.  The point I'm getting at is, yes, the clockspeed will increase without crashing, but it's just going to cause throttling and stutter by doing so, and to me, it seems like the FTW card would only run well at 1342 or so, maybe even less, max boost speed with it's current TDP limit.
 
Currently, the only card that doesn't throttle over 1500mhz is the Gigabyte.  This EVGA FTW I own throttles at stock clocks resulting in more stutter than other 970s and GPU utilization issues I was showing in the charts.

Folding is kind of irrelevant to compare to in this case.  I don't have folding experience, but I'm pretty big in the Bitcoin scene.  Typically any GPU compute application that doesn't stress memory bandwidth is only going to pull 2/3rds max of the card's power.  A game with a heavy engine or something like Furmark will blow it away in power consumption.  The only exception to this is memory hardened algorithms such as Scrypt.  In a non-memory hardened algorithm, you can overclock much higher than in an actual game with no temperature or power problems.

r0ach, there are no current plans to release an updated BIOS for the 970FTW, but as EVGATech_JaesonW had stated about his, I have also seen several of these cards clock substantially higher than yours without throttling issues.  At this point, it appears that you may have a faulty card and I would recommend doing an RMA on the card, either though EVGA or through your place of purchase if this was purchased elsewhere.

EVGATech_DaveB
r0ach, there are no current plans to release an updated BIOS for the 970FTW, but as EVGATech_JaesonW had stated about his, I have also seen several of these cards clock substantially higher than yours without throttling issues.  At this point, it appears that you may have a faulty card and I would recommend doing an RMA on the card, either though EVGA or through your place of purchase if this was purchased elsewhere.

I'm sorry but that statement about people overclocking these cards to 1500 without throttling is completely impossible when they all already throttle at stock speeds and increasing TDP by the max of 10% doesn't even fix stock clocks.  Please do not flat out lie to me.  My card functions virtually the exact same way the 970 FTW does in the Anandtech review.  The only difference is that I thought that was an early engineering sample that would have the TDP issue fixed.
 
"Despite not being temperature limited, what we can see right away is that regardless of the clock speed settings it uses, the GTX 970 FTW is TDP limited under all scenarios. At no point in time are we able to maintain the card’s top boost bin, and instead the card spends its time fluctuating between the boost bins it can hold while maintaining power consumption of 145W. The actual drop off from the maximum boost bin depends heavily on the game; some games average clock speeds close to the maximum, while others have to pull way back.
 
Otherwise we find that in the FTW configuration the card is on average clocked 138MHz higher than the reference configuration, which isn’t quite as high as the 166-189MHz difference in their official specifications. In other words, the FTW’s performance advantage is equivalent to a 140MHz overclock."
 
 
 
I've been building PCs since the 486.  I've owned most Nvidia cards since the original Geforce 256.  My guide on overclock.net on how to configure a PC to remove input lag and stutter has more views than some companies have for this entire generation of their hardware:  Google:  overclock.net: gaming and mouse response bios optimization guide for modern pc hardware 2014 r0ach edition
 
Computer graphics, hardware, software, ESreality, Bitcoin, I can walk through any of the tech burroughs and I'm known, forget about it.  In a year you will probably see Linus Tech tips replaced with r0ach's tech bunker.  Google my name.  There's me in the newspaper for cryptocurrency:   Google:  International Business Times,  Vertcoin: The Soaring Cryptocurrency Set to Surpass Bitcoin
 
I've been debugging ManuelG's screwed up Nvidia drivers for years ever since Nvidia drivers went from perfect (186.82), to bad (starting in 190 series).  If there's anyone on this planet who can tell you if constant TDP pinging is going to be a problem or not for creating tearing and stutter, that person is me.  I mean isn't it common sense?  Ask anyone with a remote knowledge of graphics if that can cause a problem and they will tell you, "probably, yes".
 
I've gone and actually tested the difference between the EVGA 970 FTW card and the Gigabyte 970 windforce.  The Gigabyte has no TDP pinging issue and when you launch the game I tested with, Lichdom, it holds a constant 1342 clockspeed with no fluctuation, a solid 99% gpu usage, and zero tearing or stutter.  The EVGA 970 FTW can't hold a constant clock speed, and it also can't hold 99% gpu utilization either at stock clocks.  It's a case of lots of tearing and stutter on the EVGA card vs zero tearing and stutter on the Gigabyte since they didn't TDP limit their cards so hard.
 
I've also recently owned twelve r9 290 AMD cards.  AMD boost does not have issues like this.  No matter how crappy of a bin the card is, it still holds a constant clock speed in games without constantly fluctuating all over the place causing GPU utilization issues and tearing/stutter.  You can try to claim to me that Nvidia says it's "ok" for things to happen like this, but just because Nvidia doesn't care, does that mean EVGA shouldn't and release cards that do not perform as smoothly in games as other companies?
 
You can easily fix this issue by increasing the 970 FTW TDP to be 20% higher than it's current setting, and limiting highest boost bin to either 1340s or 1390s.  The card has no business whatsoever attempting to boost higher into the 1400's with the insanely low TDP limit, but it still attempts to and then stutters anyway.  Even in a low power game where it appears the card is not going to have TDP issues, the clock speed will still rapidly fluctuate from 1390s to 1405 every single polling interval.  The card needs to be engineered in a more solid manner and not completely erratic.  Although Nvidia GPU boost is designed poorly, Gigabyte got it to work fine and actually hold clock speeds and maintain GPU utilization by just raising the TDP limits.
 
If you just want to get rid of me, then charge my credit card for a base clock 980 GTX ACX 2.0 that doesn't try to boost higher than it's TDP limits allow, and then refund me the price of the 970 FTW when I send it to you.  Otherwise, I'm going to be on here posting the next two years straight until they engineer this BIOS correctly to perform properly in games instead of trying to get 1% higher in benchmarks.

Azuroth
Of course, Anandtech got theirs to 1455, but I'm sure they are flat out lying as well.
 
http :// /show/8568/the-geforce-gtx-970-review-feat-evga/16

Is your post a joke?  You saw in their review how they say the card is constantly hard throttling from TDP issues.  They ran 0 tests to see if those TDP issues caused any negative effects to game fluidity, then raise the clockspeed higher even though it was already throttling to hell and back, and you call that a "successful overclock"?
 
That's like claiming setting memory to 9000 and getting green spots all over the screen without the card physically crashing is a successful overclock.
 
I actually did tests on the effect of the TDP issue by testing with this card plus the Gigabyte 970, which doesn't have TDP issues, and it's easy to tell the TDP issue causes high tearing and stutter on this card.  Overclocking a card that's already tearing and stuttering more than normal is not helping anything.

EVGATech_JaesonW

 
Here's my clocks on a quick 3 minute Furmark run. The core clock occasionally fluctuates from 1493 to 1506, but in games I have no stutter or tearing. I sincerely think it's your card at this point as DaveB said, and highly suggest you either contact us for an RMA, or pursue one through your place of purchase if you prefer.

Ummmm, I haven't seen anyone use Furmark to bench card stability or anything like that in a long time because I'm pretty sure Nvidia stuck all kinds of safeguards in the drivers to prevent people from blowing up their cards with it.  I don't use EVGA Precision.  Where's the meters showing GPU utilization, TDP, and everything else?  This screenshot could be 1493mhz at 10% gpu utilization...
 
You need to run a real, actual game that is known for being able to maintain 99% gpu utilization (ie: not Shadow Warrior) and show an Afterburner graph of clock speed over time, GPU utilization over time, and TDP % over time.  If TDP is constantly going over 100, or 110 if you have the slider at 110, then it's going to be having more tearing and stutter than it should.
 
You should really be doing the test only at stock clocks and stock power slider too, because that was my entire post is about.

Sajin
EVGATech running furmark? o_O

While I wouldn't recommend a several hour burn in with Furmark, a couple minutes isn't going to hurt a Kepler or Maxwell card, and it was the quickest, easiest way to show a high load. Absolutely don't try this with a Fermi card or earlier though, it's too risky without the power target and temp target limits of a GPU Boost capable card. It also makes a point as to the faith I have in these cards to handle even Furmark without issue, at least for a short time =)