Kepler Nvidia GeForce GTX 780

[DilTech]

If Charlie says that, then Kepler is simply going to be great. Years this guys job has been to utterly troll, flame and hate nVidia. If he is positive, this is going to be great.
That, or all his AMD contacts have been laid off and trying to find favour with the green team again lol.

That 104 is strange though, keeps popping up, the 4 has always been mid range performance hasn't it?

That's had me confused as well. Only possible explanation I can think of is that NVidia's midrange can tackle the 7970 so they feel there's no reason to bring forth the big boy until a later time. This is obviously advantageous as it gives NVidia time to tweak their design on the 28nm process before putting out a behemoth of a chip while also giving nvidia a haymaker to throw later on this year that AMD is going to have a tough time responding to. After all, as a publicly traded company, your main mission is to make the investors happy. Bringing out your $250 card at $400 will DEFINITELY make the investors happy.

At the same time that sucks for the consumer, because we'll end up paying higher prices from both camps this round in each respective segment. Right now people are blaming NVidia for AMD's pricing--but AMD's pricing on their cards is going to also bring up NVidia's pricing since they see they're able to get away with it. That's a bad thing for all of us.

This is, however, the first time I can think of that charlie has EVER been positive regarding a NVidia product. That alone gives me a lot of faith in this chip.

[El Mano]

Wow I never thought I would ever read such an statement from Charlie. Kepler is going to kick ass, head, teeth and everything.

This guy might just have stopped AMD's sales on it's track with just this single post.

[Dimitriman]

... while also giving nvidia a haymaker to throw later on this year that AMD is going to have a tough time responding to...

No...
...7990?

[GoldenTiger]

No...
...7990?

Two gpu ....

[GoldenTiger]

Do some of us have to post all those links with his comments about Cayman and Fermi again?

Bad memory I guess then, haha on my part.....
EDIT: I meant as far as dates, primarily, though is what I had remembered.

[Tha Last Meal]

I found this picture from zol site
http://diybbs.zol.com.cn/56/231_558125_49.html

The man says:
The first image GTX680!!
???

[SKYMTL]

No...
...7990?

Ummmm....no.

As stated above and many times in this forum, I'll take a $600 single GPU card over a $750 dual GPU card any day.

[Dimitriman]

Ummmm....no.

As stated above and many times in this forum, I'll take a $600 single GPU card over a $750 dual GPU card any day.

A personal choice... doesn`t take away that 7990 will likely be the fastest thing you can plug in a PCI slot..
BTW I too would make the same choice as you.. but I am talking about facts. AMD will have an answer to Kepler in the form of 7990.

[SKYMTL]

A personal choice... doesn`t take away that 7990 will likely be the fastest thing you can plug in a PCI slot..
BTW I too would make the same choice as you.. but I am talking about facts. AMD will have an answer to Kepler in the form of 7990.

If it takes a HD 7990 to truly respond to Kepler, then NVIDIA could easily respond with a dual Kepler card. Provided TDP stays within certain boundaries of course...

[SKYMTL]

A personal choice... doesn`t take away that 7990 will likely be the fastest thing you can plug in a PCI slot..
BTW I too would make the same choice as you.. but I am talking about facts. AMD will have an answer to Kepler in the form of 7990.

If it takes a HD 7990 to truly respond to Kepler, then NVIDIA could easily respond with a dual Kepler card. Provided TDP stays within certain boundaries of course...

[Dimitriman]

If it takes a HD 7990 to truly respond to Kepler, then NVIDIA could easily respond with a dual Kepler card. Provided TDP stays within certain boundaries of course...

Well you will have info way before I will so give us a hint when you know more about kepler

[DilTech]

A personal choice... doesn`t take away that 7990 will likely be the fastest thing you can plug in a PCI slot..
BTW I too would make the same choice as you.. but I am talking about facts. AMD will have an answer to Kepler in the form of 7990.

If the GK104 is faster than the 7970, and the GK104 is the midrange kepler chip, then how on earth would the 7990 have anything close to a prayer at being faster than NVidia's top end chip? Even if it DID happen, NVidia could release a dually of there own and out-right slaughter it.

If the news is true then NVidia will be in a very interesting position.

[Lanek]

I found this picture from zol site
http://diybbs.zol.com.cn/56/231_558125_49.html

The man says:
The first image GTX680!!
???

Look like a fake, since when they put the SKU numbers in bold.... can be just "thermal paste" traces, but it is a bit strange it appears specially on the "GK" and "2", look like the GK110 "fake" shown have just been modified with the 2. ( after the A2 revision surface. )

[El Mano]

There might be a 680 close but it's yet to be seen if it's really going to be a high end Kepler part or the same as 5870->6870.

[TRANCEFORMER]

first page updated. thx

GK104 impressively doubles up on GF114's core count to 768 CUDA cores - although Kepler shaders will be different from Fermi counterparts. Single precision performance is rated at above 2 Teraflops, twice that of GTX 560 Ti and over 50% higher than GTX 580. GK104 will continue to feature a 256-bit memory interface, but with frame buffer doubled to 2GB presumably at higher clocks. With everything increasing, unfortunately so does power consumption, to a TDP of 225W. A GTX 660 Ti variant is also expected in the future.

http://vr-zone.com/articles/report-n...led/14589.html

[p3gaz_001]

woooooow.... gtx 660 battles with 7970? did i read well??? ..... i think i'll be scared about gtx 680 or 780 or what ever comes out after the 660.

[ClockZ]

AMD put out a hell of a card in the 7970 IMO, if any of this speculation about how fast Kepler will be ends up being true then I will be seriously impressed and it will be huge leap over previous gen. Think I will hang onto the 480's till the dust settles but would love to have a single GPU to replace them.

[ichime]

Don't know which is causes me to scratch my head more; nVidia's mid range chip potentially being faster than AMD's high end chip or nVidia's mid-range chip potentially having a higher TDP than AMD's high end chip...

[Lanek]

Dunno, but i have the strange feeling, something still going strange in the infos we have: Why speak so much about the GK104, if the High-end will be outrageously fast ? Since when Nvidia launch a series by push first the middle range ?

[p3gaz_001]

Time will tell us what nvidia is deserving for us.... At the same time i do not want to remain dissapointed...

[hurleybird]

AMD set a pretty low bar with 7970. They can definitely raise it higher by tweaking the cooling/voltages/clocks though. A hypothetical 1100-1200MHz Tahiti seems like it would be hard to beat with GK104.

[Lanek]

Leaks speak about a 225W card, if really this is the case for the middle end cards ( GTX560 = 150W, TI= 170W, 460= 160W, GTX580 = 243W ), i fear they can't release a higher end models under 280-290W ( and this is the minimum ( expecting it have at least 1024SP), a wider memory bus and more GDDR. Untill offcourse a lot of "informations/leaks" are merged between different SKU...

[jaredpace]

Leaks speak about a 225W card, if really this is the case for the middle end cards ( GTX560 = 150W, TI= 170W, 460= 160W, GTX580 = 243W ), i fear they can't release a higher end models under 280-290W ( and this is the minimum ( expecting it have at least 1024SP), a wider memory bus and more GDDR. Untill offcourse a lot of "informations/leaks" are merged between different SKU...

That rumor bit at chiphell about the 225w GK104 suggests that Nvidia is pushing to increase PCI-E power specs to 400 watts per slot. I guess they'll need like ~300-350Watts for a decently clocked 1024SP 28nm Kepler card.

[jaredpace]

Dp.

[boxleitnerb]

You guys realize that someone probably confused the 2x6pin requirement with the TDP, right? Actual TDP will be lower, I'm certain of it.