Tuesday, 21 September 2010 18:50

After Fermi in 2H 2011 comes 28nm Kepler

Written by Fuad Abazovic
nvidia

GTC 2010: Maxwell in 2013

Nvidia just showed that after Fermi comes Kepler. This is a 28nm new processor that should get three to four times faster than Fermi. Nvidia hopes to achieve 4 to 6 Dual precision Gflops per Watt, while Fermi is less than 2 DP Gflops performance per watt.

Kepler comes in 2011 and its a 28nm part and goes in production next year.  In 2013 Nvida hopes to introduce Maxwell, a 16 times improvement in performance for parallel computing.

Kepler comes from Johanes Kepler German scientist that worked at Univesity of Graz, who after Kopernik spend a lot of time researching on astrology.

We just got an update from jensen where the chap said that Kepler comes aligned with 28nm, so in the second part of the year. Yes the big boss said that Kepler comes in second half of 2011.

gtc_roadmap_1
Last modified on Wednesday, 22 September 2010 18:04
blog comments powered by Disqus

Comments  

 
+16 #1 Jigar 2010-09-21 19:00
Nice, good for us. :-)
 
 
+56 #2 yourma2000 2010-09-21 19:17
I don't know how Nvidia can predict performance per watt of non existent GPUs, especially after Fermi poor power consumption!
 
 
+52 #3 fudzillalololo 2010-09-21 19:17
LOL, Fermi says 2009, when it actually was released in 2010.

So when can we expect Kepler? 2012?

Will it be another fail?

In 2012, nVidia will probably catch up with AMDs 2009 Evergreen performance/watt offerings.
 
 
+49 #4 fudzillalololo 2010-09-21 19:22
Quoting yourma2000:
I don't know how Nvidia can predict performance per watt of non existent GPUs, especially after Fermi poor power consumption!


Considering that nv was moaning all over before fermi release that it will blow ati offerings away and that amd should just lower their financial expectations, they are not very good in predictions.
 
 
+10 #5 boobster 2010-09-21 19:45
Quoting yourma2000:
I don't know how Nvidia can predict performance per watt of non existent GPUs, especially after Fermi poor power consumption!



Probably with computer simulations. Not 100% accurate, but better than any random person posting on the internet could guess...
 
 
+32 #6 NickThePrick 2010-09-21 20:32
2nd half of 2011 the man says ? That means we get it precisely in February 2012
 
 
+13 #7 Greg4422 2010-09-21 21:22
Kepler? Don't they still need to fix the hot, power hungry Fermi architecture? oh nevermind this is just hype in a futile attempt to stay relevant.
 
 
+18 #8 spede 2010-09-21 21:59
"Yes the big boss said that Kepler comes in second half of 2011". I think he made similar statements about Fermi schedule. Let's just add six month to the plans. Then we're ok. :)
 
 
+21 #9 Dreay 2010-09-21 22:53
How does 3-4 times the preformance per watt promise 3-4 times the overall preformance? Seriously get your shit together
 
 
-36 #10 Squall_Leonhart 2010-09-21 23:18
Fermi's late release can be entirely pinned on TMSC's poor 40nm process.
 

To be able to post comments please log-in with Disqus

 

Facebook activity

Latest Commented Articles

Recent Comments