NVIDIA Kepler speculation thread

[trinibwoy]

There is the problem. You are basing your point around Tahiti. Nvidia will also be releasing a single massive sized GPU for the Tesla/HPC Market. That GPU will also be Nvidia's high-end and will crush Tahiti in performance. So GK104 is not Nvidia's high-end whereas AMD's Tahiti is.


To quote a famous US President, "There you go again". Nvidia will not price the GK104 at Tahiti's high price even if the performance is nearly the same. The high-end price for Nvidia will be for the GK110/112.

I'll go out on a limb here and assume you don't know anything about GK110 or its release schedule. There's currently no reason to believe TSMC can supply enough 28nm volumes for a cheap 680.

Also, there's a single market and a single definition of "high-end". It might change in a couple weeks but right now that's Tahiti, end of story.

 

[AnarchX]

redquasar updated GTX 670 Ti info:
http://www.redquasar.com/forum.php?mod=viewthread&tid=7234

Something about 5% behind 7950, same SP number as 680 and a 40% bios update.

 

[SimBy]

redquasar updated GTX 670 Ti info:
http://www.redquasar.com/forum.php?mod=viewthread&tid=7234

Something about 5% behind 7950, same SP number as 680 and a 40% bios update.

40% bios update?

 

[AnarchX]

Seems to be a modificated bios. Maybe with >200W power limit?

 

[Kaotik]

Regardless of what the BIOS does, it's simply impossible that mere BIOS update could push it 40% from "5% slower than 7950". Maybe 40% from some earlier BIOS when it lagged more behind.

 

[Man from Atlantis]

redquasar updated GTX 670 Ti info:
http://www.redquasar.com/forum.php?mod=viewthread&tid=7234

Something about 5% behind 7950, same SP number as 680 and a 40% bios update.

that is what i posted a microblogging post here tomorrow which led me to redquasar forum and found 670ti pics lol.. but looks like admin and he is friends.. and PHK was walking about the same person(a.k.a. SpeedTime-closure Ke) had ''shielded version'' and actually 680 is faster than 7970..

 

[A1xLLcqAgt0qc2RyMz0y]

You really think nVidia will price GK104 low just because GK110 is coming in something like 4-6 months from now (being positive)?

Yes

GK112 or whatever the dual GK104 is

Nvidia will not release a dual GK104. The GK110/112 Single GPU will be all that is needed to take the high-end.

 

[Ailuros]

Nvidia will not release a dual GK104.

I wouldn't place any bets on that one if I were you.

The GK110/112 Single GPU will be all that is needed to take the high-end.

It's GK110 and it's not going to appear all that soon it seems, at least not for the desktop.

 

[iMacmatician]

redquasar updated GTX 670 Ti info:
http://www.redquasar.com/forum.php?mod=viewthread&tid=7234

Something about 5% behind 7950, same SP number as 680 and a 40% bios update.

Interesting… SemiAccurate claimed "8 group" vs. "7 group" GK104s and now here's a claim of 670 Ti CCs = 680 CCs.

Besides the possibility that one source or the other is incorrect, maybe that gives the possibility of 3 models: a 670 with fewer CCs and a lower clock, a 670 Ti with all the CCs and a lower clock, and a 680 with all the CCs (I'm assuming it's all the CCs…) and a higher clock? Two variants of an NVIDIA chip of ~300 mm^2 or larger with full CC counts are rare, the 8800 GTX/Ultra and the GTX 275/285 are the only ones I can find since the 8000 series and in either case, one chip launched some time before the other. If the link in the info is true, then I'm guessing that the 670 Ti is more of a "standard" clocked GK104 and the 680 is to try to beat Tahiti at least until GK110 comes along.

If the discrepancies in supposed clock speeds (705/950) weren't explained by the dynamic clocking, the above could explain that too (although the difference is quite large…).

 

[A1xLLcqAgt0qc2RyMz0y]

It's GK110 and it's not going to appear all that soon it seems, at least not for the desktop.

And it's not needed at this time as the GK104 will hold the high-end for now.

Nvidia will stockpile the GK110 for HPC/Tesla and when/if AMD releases a Dual-GPU card the GTX690 Single GPU GK110 will be there to regain the high-end.

 

[Ailuros]

And it's not needed at this time as the GK104 will hold the high-end for now.

Hold my fair warning in the back of your mind just in case. Unless of course you like rendering in absolutes for things you don't know first hand. Frankly I don't either, but then again I prefer to leave a window of doubt open just in case.

Nvidia will stockpile the GK110 for HPC/Tesla and when/if AMD releases a Dual-GPU card the GTX690 Single GPU GK110 will be there to regain the high-end.

Another world exclusive at B3D: the GK110 top dog is going to be named GTX690, it's first salvage part GTX689, the second GTX688, because GK110 is just a larger GK104 with HPC capabilities? Right?

 

[Vardant]

redquasar updated GTX 670 Ti info:
http://www.redquasar.com/forum.php?mod=viewthread&tid=7234

Something about 5% behind 7950, same SP number as 680 and a 40% bios update.

It's 5% behind 7970.

What interests me most though is the die size. Latest info suggests it's just a bit under 300mm^2.

 

[jimbo75]

Nvidia will stockpile the GK110 for HPC/Tesla and when/if AMD releases a Dual-GPU card the GTX690 Single GPU GK110 will be there to regain the high-end.

When was the last time Nvidia's top gpu beat AMD's dual?

 

[UniversalTruth]

Ok, fair enough. Let's say Tahiti is overpriced so GK104 has to earn its stripes against GF114. Current rumors put it at 70-90% faster than the 560 Ti in BF3 with a smaller die size....would that be impressive enough?

Yes that's definitely more impressive than the 45-50% gain we see with Pitcairn over Barts. I'll wait until the numbers arrive, to be sure.

Absolutely amazing it would be.
The jump from Cayman to Tahiti is the same (or slightly worse) as the one from Barts to Pitcairn - around 40%..
My question is- if GK104 is 70-90% faster than GF114 in BF3, why wouldn't it be on average around those numbers, I mean why the performance improvement in BF3 would be higher than the average one?
And the second question- what kind of CPU do you think won't be a bottleneck for a potential 7990 or corresponding dual GPU card from nvidia 2 X GK110? I mean TPU (just as an example) tests with i7 920 @ 3.8 GHz but...

 

[boxleitnerb]

^ that's what SGSSAA is for
If we're talking newer DX11 games that make good use of computing power I guess you could see thoes 70+% on average no problem.

 

[Man from Atlantis]

My question is- if GK104 is 70-90% faster than GF114 in BF3, why wouldn't it be on average around those numbers, I mean why the performance improvement in BF3 would be higher than the average one?

looking to 7870 reviews comparing to 7950, 7870 is too close to 7950 in BF3 .. probably it is shader heavy, not like other console ports..

 

[psurge]

AlexV - are you allowed to say anything about when people that actually know something will be allowed to say anything? Hitting refresh on this thread is tiring out my index finger.

 

[Man from Atlantis]

Do you think this has anything about Kepler? i see 4 MCs and PCI3 I/O and other stuff

Spoiler

http://www.gputechconf.com/page/sessions.html

 

[Mize]

AlexV - are you allowed to say anything about when people that actually know something will be allowed to say anything? Hitting refresh on this thread is tiring out my index finger.

Lol. This.

 

[fellix]

Do you think this has anything about Kepler? i see 4 MCs and PCI3 I/O and other stuff

That's the top metal layer, I think. Not much to see really - just the overall layout.

p.s.: By the I/O pad population, this chip seems to be a 256-bit device.