News : Nvidia’s 28nm mobile lineup leaked

nytmode

Distinguished
Mar 20, 2009
18
0
18,510
What does Nvidia (NASDAQ:NVDA) have coming up for the post-Christmas GPU line? You have heard a lot about the 28nm parts, and here is what you will be seeing.

The short story is this, Nvidia is putting out a bunch of Fermi shrinks on 28nm, and you will likely see the mobile variants first. They are as follows, with some information a bit blurred to protect the exact sub-species of mole involved.

Nvidia_mobile_GPUs.png


A few things to note here, most notably that this is NOT the complete lineup, there are very likely to be a few 40nm parts here and there to fill in gaps. If you are paying attention to the details, you will notice that the first samples have been delayed about two months. This means a big miss on the “non-denominational winter holiday that neither affirms or denies any particular faith but allows everyone to celebrate as (bored) equals”(TM) selling season.

If you recall SemiAccurate’s analysis around Kepler’s release, this adds a bit to the mix. Short story, any Kepler based parts will trail the above by at least three months, likely more. If the first Nvidia 28nm parts are not due to market until January, then Kepler will be almost impossible to get out in Q1. This should shock no one.S|A

by Charlie Demerjian

Source: Semi-Accurate
 

benski

Distinguished
Jun 24, 2010
1,611
0
19,960
"and here is what you will be seeing."

What is it that we are seeing? Projected 3DMark scores (no indication as to version), estimated TDP, and not a whole hell of a lot else, and then some speculation on Keplar's release with no substantiation?
 
Low TDPs and high memory bandwidth tells me those numbers got pulled out of a dark hole and are pretty well make believe.

nVidia wouldnt release something with a TDP of 75W max and a 256b memory bus, then what do they have to offer their higher end? If they got their good chips down anywhere near that low that would show a massive archtecture and marketing shift for nVidia, so if those numbers are real, they arent the chips that many of us are interested in.
 

If they use slower GDDR5, and their high end is 384 or better, I can see these asmid low end to low end