Nvidia officially green lights custom designed Titan boards

[blackened23]

Actually I love custom cards. You took everything I said the wrong way. I just think because the card is so expensive to start with $200 isn't a reach. Take a look at the EVGA SC Signature. It's already $70 more than a regular Titan.

I believe Kepler might be a bit sensitive to voltage because of the limits NV is putting on it right now and the AIBs are going to want to make sure they don't lose too much on RMAs should they occur.

The thing about the SC signature is that it isn't a custom design. It is a reference design, and the higher cost is for a mousepad and T-shirt. Yes you read that correctly, a tshirt and mousepad.

You may be right on the voltage sensitivity, i'll grant you that. On the other hand, a custom card is completely pointless unless it adds value to the original product; custom cards add value with premium components, more power phases, and more power draw (2x 8 pin connectors). And they still have to be certified by nvidia. I am confident that if custom Titan cards do come to fruition, they will have added value above reference (there is really NO POINT if they don't...).

I guess we'll see what happens. I'm a longtime fan of the MSI lightning (I also like the newer DC2 double slot designs quite a bit...) so i'm anxious to see what happens.

 

[DooKey]

snip

I guess we'll see what happens. I'm a longtime fan of the MSI lightning (I also like the newer DC2 double slot designs quite a bit...) so i'm anxious to see what happens.

I owned 2x MSI 6970 Lightnings and they were awesome cards. I also owned a 5870 Lightning so I'm well versed with them. Hopefully the Titan Lightning will be as good.

 

[3DVagabond]

Very true, assuming this story comes to fruition - i'm hoping for a MSI lightning variant with 2x 8 pin connectors and the usual array of excellent Lightning PCB components. Basically, there's no reason for a custom card if there are no special features -- I think nvidia realizes this. A lot of people including myself love custom cards and the benefits they provide over reference.

If this happens (lightning titan), count me in for the purchase of a Titan.

We have Lightning and DC II 680's. Plus a slew of others and the voltage is still locked. We'll have to see what happens with Titan custom cards.

 

[Lepton87]

It really boggles my mind that people are so vigorously defending the titan's reference PCB along with the paltry power limit of 265W. The card costs a 1000$!!!

Yes! Lets dump all that heat inside the case and good luck with SLI.

Have you already bought the reference titans? It's understandable that people want to defense their purchases, regardless it's really a narrow-minded comment. Maybe someone wants more then 265W from their cards and wants to water-cool their cards?

 

[thilanliyan]

Good to hear that there will be non-reference Titans. Hopefully there will be a decent amount of manual control.

 

[boxleitnerb]

I've not hit the 100% TDP mark on my Titans yet. They downclock due to temperature and fan activity hmm only.

 

[CurrentlyPissed]

I've not hit the 100% TDP mark on my Titans yet. They downclock due to temperature and fan activity hmm only.

Sarcasmn, right?

 

[blackened23]

Sarcasmn, right?

No, sadly. There seems to be an issue with the 314.09 drivers causing Titan cards to throttle/downclock anytime you use a manual fan setting that is too high. There's a huge thread over at OCN with callsignvega discussing it.

I'm using a modded version of 314.09 on my lightning 680s and they always downclock to 1050mhz if I use a high manual fan speed. This never happened with prior drivers.

Hopefully newer drivers will fix this.

 

[AdamK47]

Will there be aftermarket cards with blower style coolers?

 

[blackened23]

I presume that EVGA would make a classified version - all of their stuff generally has blower style coolers.

 

[BoFox]

No, sadly. There seems to be an issue with the 314.09 drivers causing Titan cards to throttle/downclock anytime you use a manual fan setting that is too high. There's a huge thread over at OCN with callsignvega discussing it.

I'm using a modded version of 314.09 on my lightning 680s and they always downclock to 1050mhz if I use a high manual fan speed. This never happened with prior drivers.

Hopefully newer drivers will fix this.

Strange that NV would do this... weird glitch.

So, you run your 680s at 1.4v? What's the normal oc'ed speed?

 

[blackened23]

Strange that NV would do this... weird glitch.

So, you run your 680s at 1.4v? What's the normal oc'ed speed?

1.4V no way! Not 24/7 anyway.....I can get 1350-1400mhz fairly effortlessly with +100mV which works out to 1.275V? Or something along those lines....1.3V is my limit for 24/7 operation.

Anyway, 314.09 drivers have a weird glitch that causes boost speeds to be affected when your manual fan setting is too high. If you leave the speed at 60% you should get the best results though - nvidia is (apparently) working on a new unified driver, i'm sure the glitch will be resolved then.

 

[2is]

Yes! Lets dump all that heat inside the case and good luck with SLI.

I've got 120+200mm exhaust fans to deal with that. Not that it matters to me as I'm not in the market for Titan, but I am interested to see what they can do without the artificial nannies in place.

 

[UaVaj]

good news - glad did not jump on the reference design right away. looks like there is some of headroom for a decent oc.

 

[3DVagabond]

good news - glad did not jump on the reference design right away. looks like there is some of headroom for a decent oc.

Depends if they can eliminate the 265W cap. If not, then super coolers are a waste. The stock cooler is quiet and has the added benefit of expelling the hot air out of the case.

 

[2is]

Depends if they can eliminate the 265W cap. If not, then super coolers are a waste. The stock cooler is quiet and has the added benefit of expelling the hot air out of the case.

That's really going to be the key thing here. Seems like it would be almost completely pointless unless that cap was removed. And if it is removed, I wonder if it would be possible to flash the bios of their reference cards with the non-reference one.

 

[UaVaj]

non reference would suggest 265w cap removed.

 

[3DVagabond]

non reference would suggest 265w cap removed.

Why do you think that? Suppose it's driver enforced? Or tied into the 2D -> 3D clock control? The AIB's might not have any control over it.

 

[wand3r3r]

non reference would suggest 265w cap removed.

Yes just like the classified and lightning 680's which are neutered along with all the rest of them? (Yeah you can flash 680's, but there goes your warranty)

NV likes to neuter their cards (and their fanbois?), I will believe the titanic is un neutered when I see it.

 

[Roadrunners]

Looking at overclocking results Titan seems to give a poor return of frames for percentage overclock. A 35% overclock seems to only give a 10% boost in frames which is very disappointing.

 

[2is]

That's because there's no such thing as a 35% overclock. The speed gets throttled as soon as power consumption approaches 265 watts.

 

[Roadrunners]

That's because there's no such thing as a 35% overclock. The speed gets throttled as soon as power consumption approaches 265 watts.

Guru3d shows clocks of 1170mhz and it only it only yields 10% boost in frames.

http://www.guru3d.com/articles_pages/geforce_gtx_titan_overclock_guide,1.html

User benchmarks are also showing similar results.

 

[3DVagabond]

You also have to consider that stock the cards often boost higher than the advertised boost speeds.

 

[Leadbox]

Looking at overclocking results Titan seems to give a poor return of frames for percentage overclock. A 35% overclock seems to only give a 10% boost in frames which is very disappointing.

It's the boost thing that confuses things a bit here, Anand's titan was boosting to just under 1GHz during their benchmarking. So a 35% oc from 837Mhz gives you 1130Mhz but if you were already boosting to nearly 1000Mhz, you're really only getting ~130Mhz which lines up well with the 10% boost you're seeing. Thats how I think it works anyways:biggrin:

 

[2is]

Guru3d shows clocks of 1170mhz and it only it only yields 10% boost in frames.

http://www.guru3d.com/articles_pages/geforce_gtx_titan_overclock_guide,1.html

User benchmarks are also showing similar results.

That's because there's no such thing as a 35% overclock. The speed gets throttled as soon as power consumption approaches 265 watts.

I'm not really sure how else to word this.. Yes, you can overclock the card, but that overclock will not hold. What started out as a 35% OC ends up being far less, exactly how much less varries.