Nvidia BigK GK110 Kepler Speculation Thread

[A1xLLcqAgt0qc2RyMz0y]

"Possibly as high as 45% on average depending on the game selection" is what you were meant to read.

Well I certainly don't see how I can read that from this statement:

I guess we'll be looking at anything between 25% as a low point and possibly as high as 45% (faster than the Ghz Edition) as a high point. Pretty crazy when you think about it. 45% is reasonable and 25% is terrible, that's quite a difference.

Nor does it make any sense to have an average vary from 25% to 45%. That would be an 80% variation. In fact the above mentioned chart shows 40% average improvement for nine games vs the 7970GE.

 

[lanek]

The only few results we have seen, are all over the place. ( normal you will tell me, depend the games, res and settings ), but i have the feeling we will need take 4-5 reviews and make a good comparaison between them with it. ( why i have the feeling the 690 results are completely crippled... scaling look poor, a lot lower of what i was remember when the 690 have beeen released ).

 

[jimbo75]

Well I certainly don't see how you can get that from this statement:

Well if you had looked at the chart I posted in the same post, you'd have seen that the variation by game goes from 21.57% to 65.71%. I'm hardly about to suggest different figures when they are right in front of me. Use a bit of common sense - of course I meant on average.

Nor does it make any sense to have an average vary from 25% to 45%. That would be an 80% variation.

And that's what we'll end up with, of that I'm sure. We've had 27% low point (HWC) and 40% high point (Guru3d) already, and I don't have any doubts that they won't be the furthest outliers today. I have a feeling Hexus will get 40%+ faster for Titan anyway.

 

[UniversalTruth]

Draw distance and shadows have always been the killer for cpu's. I'm unsure what AMD has done to make the FX perform so well but tbh...I wouldn't be surprised if there were some shenanigans involved I don't really get how the 6 and 4 core FX's are performing so well otherwise.

Yes, there are shenanigans involved. But no, they are not coming from AMD. The wintel duopoly is doing it. I hope now things will change with this very important console market win for AMD.

 

[Alexko]

Well I certainly don't see how I can read that from this statement:

Nor does it make any sense to have an average vary from 25% to 45%. That would be an 80% variation. In fact the above mentioned chart shows 40% average improvement for nine games vs the 7970GE.

Yes but this one shows 27%:

So a 25~45% range is actually fairly reasonable.

Edit: oops, too late.

 

[lanek]

its open bar: could find the link of the review grouped here in case..

http://www.xtremesystems.org/forums...ITAN-reviews&p=5172283&viewfull=1#post5172283


http://www.techpowerup.com/reviews/N...rce_GTX_Titan/
http://www.techpowerup.com/reviews/N...GTX_Titan_SLI/
http://www.guru3d.com/articles_pages..._review,1.html
http://www.guru3d.com/articles_pages..._review,1.html
http://www.overclockers.com/nvidia-g...eo-card-review
http://hexus.net/tech/reviews/graphi...rce-gtx-titan/

http://www.pcper.com/reviews/Graphic...-Rating-Update
http://www.anandtech.com/show/6774/n...mance-unveiled


http://www.hardwarecanucks.com/foru...geforce-gtx-titan-6gb-performance-review.html

( Its not me who research the link, so ofc, thanks to peoples who have do it on xs )

Like i was suspect it, some dont use the last beta for AMD.

 

[A1xLLcqAgt0qc2RyMz0y]

The only few results we have seen, are all over the place. ( normal you will tell me, depend the games, res and settings ), but i have the feeling we will need take 4-5 reviews and make a good comparaison between them with it. ( why i have the feeling the 690 results are completely crippled... scaling look poor, a lot lower of what i was remember when the 690 have beeen released ).

If the final reviews do result with huge variations (for the same games at the same settings) then it is possible that some of the reviews are CPU bound and/or not using the release drivers.

 

[jimbo75]

Well I was completely wrong about Hexus - their 5 game suite ended up with a miserable 15% lead for Titan (lost in Showdown and Sleeping Dogs). They still gave it 4 stars though, I fear that decision will come back to haunt them today

32% on Techpowerup.

551mm2? Jesus.

 

[lanek]

i have only watch 2-3 reviews, but what surprise me is despite the impressive performance of Titan, how it show how not so bad is the 7970GE.. ( i ask me if AMD could add 5-10% more with some future drivers on some title )

( interesting to see Anandtech reviews for computing too )

 

[CarstenS]

FWIW - we don't have the most cards nor the most benchmarks, but we tried to put GPU-Boost 2.0 to more use cases. Also, having the card operate under 28°C air temperature to see real world performance.
http://www.pcgameshardware.de/Grafikkarten-Hardware-97980/Tests/Test-Geforce-GTX-Titan-1056659/

 

[lanek]

FWIW - we don't have the most cards nor the most benchmarks, but we tried to put GPU-Boost 2.0 to more use cases. Also, having the card operate under 28°C air temperature to see real world performance.
http://www.pcgameshardware.de/Grafikkarten-Hardware-97980/Tests/Test-Geforce-GTX-Titan-1056659/

Thanks for the link, i offtly forget this site lol . really interesting to see the Boost 2.0 tested this way,

 

[DSC]

http://www.geforce.com/whats-new/articles/introducing-the-geforce-gtx-titan

 

[Alexko]

http://www.hardware.fr/articles/887-27/recapitulatif-performances.html

Damien found 12~23% on average over the 7970 GHz Edition, depending on definition and Boost settings.

He also found that repeating the same benchmark twice tends to lead to lower scores on Titan, because its Boost is temperature-dependent. For example: in Anno 2070: 75 FPS on the first run, 63 FPS with a warm card. That's a pretty massive difference, and I wonder how many reviewers even noticed.

 

[CarstenS]

Thanks for the link, i offtly forget this site lol . really interesting to see the Boost 2.0 tested this way,

Thanks for the laurels. Please note, that with the 28°-C-Setting, what we did was not only to use this temperature, but we've logged and later forced, the lowest clocks for each game/resolution combination, in order to provide a next-to-real-world case. Even short loading sequences before a benchmark run can cool down the GPU to a point, where 100 MHz Boost is applied over what is achieved in continous gameplay.

Just in case the translation did not make this clear.

 

[jimbo75]

http://www.hardware.fr/articles/887-27/recapitulatif-performances.html

Damien found 12~23% on average over the 7970 GHz Edition, depending on definition and Boost settings.

He also found that repeating the same benchmark twice tends to lead to lower scores on Titan, because its Boost is temperature-dependent. For example: in Anno 2070: 75 FPS on the first run, 63 FPS with a warm card. That's a pretty massive difference, and I wonder how many reviewers even noticed.

I have a real appreciation for Hardware.fr reviews. And you're right, that's an incredible difference in scoring between the hot and cold card, and is something that should be brought to the attention of all reviewers.

 

[xDxD]

Too much money per performance. Sorry nvidia, it's a fail at that mrsp. Imho the right price to be attractive is no more than 750$

 

[jimbo75]

Gotta say power consumption is pretty spectacular. Looks about even or even slightly better than the 7970 GHz.

 

[elect]

Disappointed for the gpgpu performances

http://www.tomshardware.com/reviews/geforce-gtx-titan-performance-review,3442-10.html

 

[lanek]

Gotta say power consumption is pretty spectacular. Looks about even or even slightly better than the 7970 GHz.

Without saying, the quality of the cooler and design of it is excellent ( sound and cooling wise ).. I hope it could become a standard for next AMD and Nvidia cards.

 

[trinibwoy]

Gotta say power consumption is pretty spectacular. Looks about even or even slightly better than the 7970 GHz.

Hardware.fr must have some sort of golden sample. It consumed less than a 680 in one test. Overall, nice performance bump over the 680, excellent power consumption but prohibitive pricing.