Quick question. I know some guys have put EK 780 Classified block on their 980 Ti Classified. How did you manage to keep the original 980 Ti Classified backplate. Can you please link me to the screws I need for it? Thanks.

Can somebody please give me some settings for a gentle overclock. I have an EVGA Classified. I've looked around for instructions but cannot fathom it. Using MSI Afterburner and just do not understand what to do!!

ericbartman

Mangofire
This is why we Classify! I would expect the same out of the ti's. Would some company "EK" hint hint please make a water block soon for da twins.
 

I am happy to see someone pushing 1600 MHz, because for me it has done nothing but disappoint. As a background, I replaced my 980Ti SC+ with an everyday BIOS (using Sajin Mod) overclock of 1484 (65.5% ASIC), with a 980Ti "Classified" which is unstable at 1468Mhz (77.9% ASIC), and crashes every 10-15 minutes or so.
 
In search for information, or see what I might be doing wrong, I have researched and tweaked and tuned for over 18 hours now, still no luck. Frankly, I have given up at this point. If things don't improve, basically, this will be my first and last classified card.
 
EDIT: I just had a crash at 1450MHz. At 77.9% ASIC, this is 100% pure crap.
EDIT2: I just crashed at 1440Mhz on Valley.
 
 
Sajin! Please help!!!

What driver you running? I had a 980 from another vendor and as soon as the witcher 3 driver dropped I couldn't overclock the card at all! The card was stable for about 3 months before that. So maybe try another driver. Hope that helps as I have just bought a classy! Not sure what I would do in your shoes. Mine boosts to 1405 out of the box so if it only went that high I might look at my options!

AMP_US
First, ASIC has almost no bearing on overclocking on air. Second, people need to understand that Maxwell is HIGHLY susceptible to the silicon lottery. Higher clocks only come with lower temps, not more voltage. If you got a stinker... no custom PCB or BIOS tweaking or 1.3V is going to save you. I have seen golden reference cards (on water) out perform KPEs. In addition, this difference in clock rarely equates to noticeable FPS differences (unless you really get a crappy card) so the only thing you are missing out on are benchmark leaderboards.
 
Look at MSI's "Lightning"... same deal.
 
For Pascal, I really hope EVGA focuses on two things for their Classified cards, out of the box cooling and making them feel special.
 
Nvidia is likely going to continue this "energy efficient" design route (b/c w/large scale compute business it makes sense). I would not be surprised if this gives us a similar situation. The Hybrid (AIOs from other brands as well) was obviously a massive success and I think that AIO style card should carry over the the Classy and KPE out of the box. A 1080 Ti Classified with an expandable (important for SLI and case rad options) 140-240mm AIO with quality components (dual high SP fans, thicker/denser rad, full cover copper water block) would be an interesting direction. You can still have a Hybrid (reference card w/ 120mm AIO) but give the Classy/KPE the best possible shot out of the box (outside of LN2). Hopefully the smaller form factor for Pascal allows for some more interesting form factors (single slot w/ water pump on card).
 
The KPE is pretty "special" but IMO the Classy looks nor feels "special". I am not say the EVGA understated design or Classified color pallet should change, but the Classy should have a differentiating visual identity. Really, this is a complaint with the majority of video card designs. They are either garishly ugly or bland and uninspiring. IMO the 285 Classified was good. Asus ARES III as well.

Hi AMP, thanks for your reply. You mentioned in your reply that Thermals decide the overclock of the card and I agree with that, as it is one of the factors.
 
Now I may ask the question, how does this heat get generated to affect the thermals? Simple answer, more power pushed into the card = more heat generated, which needs to be dissipated in order to keep thermals in check, to match your first point. Agreed?
 
In detail:
Each transistor switching cycle takes certain amount of energy (this is why smaller transistors take less energy to switch and hence you get better efficiency on a smaller transistor size, as we know 28nm in Maxwell's case). Now this energy amount is fixed, does not change, and let us call it VT.
 
As we know that switches per second is your frequency, hence higher frequency = higher switches per second. Considering the above, the more switches per second you need, the more energy per second (which is power) needs to be provided to the chip. So far with me?
 
Now what provides this power to the transistor that is switching? In the simplest terms, the voltage applied across the circuit. Now this is where ASIC quality comes into play, when you have a higher ASIC quality, most of the voltage you apply (VDDC), goes to the transistor to satisfy its VT requirement. Hence, you can provide power to transistor, without having to push up VDDC too much. But if you have a lower ASIC quality, think of it as adding internal resistance, where you need to apply higher voltage to the VDDC to push the same VT. Still with me?
 
Now what is needed to maintain high clocks? Consistent VT with scaling current(I) based on frequency. Which is constant power, which why you need good power supply and high phases to regulate it better. Since, VT depends on VDDC and ASIC quality, the power that reaches the transistor depends on ASIC quality.
 
TLDR:
In summary, if you have bad ASIC quality, you will have to supply higher voltage and power to the chip, which will generate more heat, and make the chip unstable. However, when ASIC quality is good, you don't have to apply higher voltage which reduces the power requirement and heat generated, and you reach higher clocks. 
 
Since classified has higher ASIC quality (77.9 instead of 65.5 on my SC+), and bigger heatsink and fans, that is not the problem. The problem is that I am unable to control VDDC on Classified. On my earlier chip, I was able to directly control VDDC and I was able build an overclock at 1468-1481Mhz, and never had a single crash.
 
 
Edit: Engineer with ULSI specialization here. Sorry if I lectured a little.

ericbartman

AMP_US
First, ASIC has almost no bearing on overclocking on air. Second, people need to understand that Maxwell is HIGHLY susceptible to the silicon lottery. Higher clocks only come with lower temps, not more voltage. If you got a stinker... no custom PCB or BIOS tweaking or 1.3V is going to save you. I have seen golden reference cards (on water) out perform KPEs. In addition, this difference in clock rarely equates to noticeable FPS differences (unless you really get a crappy card) so the only thing you are missing out on are benchmark leaderboards.

Hi AMP, thanks for your reply. You mentioned in your reply that Thermals decide the overclock of the card and I agree with that, as it is one of the factors.
 
Now I may ask the question, how does this heat get generated to affect the thermals? Simple answer, more power pushed into the card = more heat generated, which needs to be dissipated in order to keep thermals in check, to match your first point. Agreed?
 
In detail:
Each transistor switching cycle takes certain amount of energy (this is why smaller transistors take less energy to switch and hence you get better efficiency on a smaller transistor size, as we know 28nm in Maxwell's case). Now this energy amount is fixed, does not change, and let us call it VT.
 
As we know that switches per second is your frequency, hence higher frequency = higher switches per second. Considering the above, the more switches per second you need, the more energy per second (which is power) needs to be provided to the chip. So far with me?
 
Now what provides this power to the transistor that is switching? In the simplest terms, the voltage applied across the circuit. Now this is where ASIC quality comes into play, when you have a higher ASIC quality, most of the voltage you apply (VDDC), goes to the transistor to satisfy its VT requirement. Hence, you can provide power to transistor, without having to push up VDDC too much. But if you have a lower ASIC quality, think of it as adding internal resistance, where you need to apply higher voltage to the VDDC to push the same VT. Still with me?
 
Now what is needed to maintain high clocks? Consistent VT with scaling current(I) based on frequency. Which is constant power, which why you need good power supply and high phases to regulate it better. Since, VT depends on VDDC and ASIC quality, the power that reaches the transistor depends on ASIC quality.
 
TLDR:
In summary, if you have bad ASIC quality, you will have to supply higher voltage and power to the chip, which will generate more heat, and make the chip unstable. However, when ASIC quality is good, you don't have to apply higher voltage which reduces the power requirement and heat generated, and you reach higher clocks. 
 
Since classified has higher ASIC quality (77.9 instead of 65.5 on my SC+), and bigger heatsink and fans, that is not the problem. The problem is that I am unable to control VDDC on Classified. On my earlier chip, I was able to directly control VDDC and I was able build an overclock at 1468-1481Mhz, and never had a single crash.
 
 
Edit: Engineer with ULSI specialization here. Sorry if I lectured a little.

I completely follow. That was a good explanation. Although, can't you use the Classified Voltage Tool or Precision X KBOOST to directly control/force voltage?

AMP_US
I completely follow. That was a good explanation. Although, can't you use the Classified Voltage Tool or Precision X KBOOST to directly control/force voltage?

I tried with GTX Classified control (2.1.0 and 2.1.2) and Precision X without KBoost, but it was still bouncing around. Hmm... didn't try Kboost. Doesn't Kboost fix the clock speeds at max.

I have been told that 353.62 works reliably on Win 10. I run Win 10 aswell, althgouh I am on 359.00. A few people have said that the official Win 10 drivers (355 onwards) actually made things worse for them on Win 10. Have you stopped windows from automatically updating drivers?

Both of my EVGA 980 Ti Classifieds in SLI Boosted to 1405 right out of the box via GPU Boost 2.0 in EVGA Precision x16. For most people that is a very strong overclock if you are using a reference or ACX 2.0 model. 
 
I have been able to get close to the 1.5 GHz target but I cannot sustain it. With voltage turned up and power target at max I've managed a sustained 1432 MHz for around forty-five minutes.
 
The reality is this: These cards are built and designed with outstanding craftsmanship. What they do on their own with Precision x16 and GPU Boost 2.0 is quite really enough for 98% of users. 
 
I'm currently building a custom coolant loop to dedicate to my 980 Ti Classy SLI setup. I think liquid cooling or LN2 is the only way to push much higher than 1.5 GHz. Fortunately, the Classified (and Kingpin Edition of course) beg to be put underwater despite their stellar performance on air.
 
Consider liquid cooling before you try pushing them too hard. They already run very fast as is.

Hello everyone ! 
 
I need help ..... i recently purchased a EVGA GTX 980Ti Classified .... very excited to receive the GPU , but there is one thing that i am confused about on which i need help .
 
GTX 980Ti Classified GPU needs a (2) 8 pin PCI e cables to power it up ..... my power supply is a cooler master 800W silent pro gold . Now the confusion part...... i have (1) 6 pin and (1) 6+2 pin cable coming directly from the power supply .....  CM has provided (2) 6+2 pin cables ..... but these cables have one 6 pin on other side ...
 
Can i use than cable which has 6 pin on one side and the other side has 6+2 pin to power my GPU for (2) 8 pins ?
 
Thank you .
 
Regards
Fahd. 

Fahd Chhapra
Hello everyone ! 
 
I need help ..... i recently purchased a EVGA GTX 980Ti Classified .... very excited to receive the GPU , but there is one thing that i am confused about on which i need help .
 
GTX 980Ti Classified GPU needs a (2) 8 pin PCI e cables to power it up ..... my power supply is a cooler master 800W silent pro gold . Now the confusion part...... i have (1) 6 pin and (1) 6+2 pin cable coming directly from the power supply .....  CM has provided (2) 6+2 pin cables ..... but these cables have one 6 pin on other side ...
 
Can i use than cable which has 6 pin on one side and the other side has 6+2 pin to power my GPU for (2) 8 pins ?
 
Thank you .
 
Regards
Fahd. 

HI Fahd, you should have adapters and be able to use those. Just leave the extra 6 pins out of the way and use the 6+2 portion.

Yes thats pretty normal Fahd . I have an ax1500i and each 8 pin is 6+2 . 

Drey070
Yes thats pretty normal Fahd . I have an ax1500i and each 8 pin is 6+2 . 

Thank you for the reply please see the picture that I attached it is 6 pin to 6+2 pin

http://www.3dmark.com/3dm/9904173?

My best firestrike scores at this time.

Stock ln2 BIOS. Asic 69,9%

How i can ADD more voltage because gpuz shows allways 1.2v when ADD More in PX?
Im New NVIDIA overclocker!

Hard to reach +22k Graphics score

kurosuke7t9

Razor Time
Just wondering as a guess how much high will the Classifieds overclock compared to the reference cards? Not sure if I should wait it out for the Classifieds or not? I will be placing what ever card I get on water so overclocking is fairly important to me.

 
Pretty interesting finding for me today though.

I had 2 980 Ti EVGA Hybrid Cards. Reference Design PCB but with the AiO Cooler from Asetek. Cost £640 at the time, and Boost to 1380MHz max, anymore and its unstable both on synth and live gameplay. ASIC's were 64% and 67%. (Its the Hybrids I got on my Mods Rigs.
then
I had 2 980 Ti Kingpins both from 74% Bracket. Real ASIC value are 74.5% and 75.9%. Cost me 1,005 Euro's Each including the German Postage + postage Tax. its equivalent to £740 at the time.
Kingpin Boost to 1430MHz out of the box no overclocking. I pushed it to 1495MHz on air and that was stable on synth benchmark's but wasn't as stable when you keep the OC and do a live gameplay. It usually down clocks to 1480MHz on live gameplay.
 
Now I got another 980 Ti Classified today, this time from Scan UK with game + free Next Day delivery. Cost £569. And I'll be damned its 78% ASIC, boost to 1452 MHz out of the box. 78% ASIC  and a higher boost clock without paying the premium from the Kingpins. Ill fiddle with its OC later when I get my used 780  classy block on it, waiting to get new rubber gasket to replace the old one's 

Welcome to the game of winning and loosing the silicon lottery. I had a Hybrid as well. Card was terrible. 1380 boost out of the box, but couldn't manage more than 1438Mhz OC'd (and good luck getting it to stay there). It also got really hot (65C). My Classified 73% ASIC, boosts to about 1418/7012Mhz out of the box. 1530/8000Mhz overclocked on air. And now 1550/8100Mhz in a custom loop. I have heard some high ASIC (>80%) Classifieds can get to 1600/8200+ (on water). Without a doubt, the Classified will go further than your average board, but the performance you experience is mostly dependent on the silicon lottery.
 
This is my shtick and I should probably just make it my signature....
 
EVGA, please make the next range of Classified cards come standard with an expandable AIO water block (Hmmm... building off of what was shown at CES this year... hmmm). Let the FTW take the custom PCB on air semi premium market and give the Classified range the design to maximize performance out of the box.

DennyDj86
Hello there! I received one gtx 980ti classified today. When i changed the thermal pad , i saw the vram chip is Samsung. My friends said to me "it isn't normal cause just Kingpin version mounts Samsung chip. " It's really?it's my first time with evga graphics card ..

EVGA recently changed to Samsung memory for the standard Classified. Lucky you!