uw buddy. more news should be out soon
Printable View
to people who believe in Tenerife, or even its demonstration in March 2012 ... are you high?
[align=center]NVIDIA Kepler GK104 (GTX 680) Final Specifications Leaked[/align]
*The site reports that Nvidia will launch a 28nm product based on the GK104 chip on 12th March and would be available globally on March 23rd.
*the GK104 would hold a total of 1536 Stream Processors, Core clock would be set at 705Mhz and Processor clock at 1411Mhz. Memory would consist of 2GB GDDR5 (256-bit) buffer clocked at 3004MHz (6.0Ghz Effective Dual Data Rate)
*that the card would either feature 2 x Six Pin or 1x 8 Pin + 1x 6 Pin PCIe connector and a die size of 320mm2 which is 45mm square smaller than the Tahiti XT chip. A 5-phase NVVDC configuration powers the GPU, The green pcb suggests that the card is in engineering phase, Final product would feature a black PCB.
How fast do you guys think GK104 will be against the GTX580?
Well, for those who know architecture leave a question. With 1536 Stream Processors (3x GF110) to 1411Mhz, 705Mhz GPU, 2GB GDDR5 (256-bit) clocked at 3004MHz buffer and a BUS 192 GB, you can expect? Working with rumors of performance that we have, you can expect something big (I mean performance) as in the G80? I think not
I donīt know how it will perform, but if Nvidia can beat Tahiti with 320mm2 chip, AMD is definitely in trouble. Perf/mm2 was their strongest point in the post-Ati generations.
The is a big catch that is left out of the leaks.
As if he's gonna say that :D
If i want the most powerful single gpu card and i have time to wait until 1st april. Should i buy 7970 now or wait for greens card?
Giving advice on this shouldnt be an NDA violate right? :rolleyes:
Maybe the big catch is Kepler has some sort of hyperthreading?
Taken from Beyond3D: http://forum.beyond3d.com/showthread...=58668&page=85
http://www.microsofttranslator.com/b...20120306077%2FQuote:
I think the details I want to tell opportunity again, NVIDIA's approach is that "to make as we think 'Hyper-Threading' Intel's seen, CUDA-Core free resources available other threading, increased performance per Core CUDA, lowered power consumption per performance to" (official NVIDIA).
That looks like LenzFire's specs.
Notice it also says GK110... anyway irrelevant, that's still not confirmed specs for that chip either, it's been said it's based on rumors.
in the news talks about the use of HT in Kepler, little consumption and performance of a 1.5 x higher than the Cayman, the words of a director of Nvidia -> http://www.microsofttranslator.com/b...20120306077%2F
It does not speak anything about cayman. Then it goes on and the big one i think is this "absolute power should be significantly improved". So first part with 1,5 times "conventional" is reletive one...
Also
Quote:
Originally Posted by Yahoo
I think we need someone native to tell us what it really says...Quote:
Originally Posted by Google
I think he says that "performance only" users wonīt be happy with that 1.5x scaling over last gen, but they will bring good power consumption improvements.
I wonder if the "hyperthreading" function being mentioned is just and update to fermi's gigathread scheduler.