NVIDIA reveals Fermi's successor: Kepler at 28nm in 2011, Maxwell in 2013

Page 7 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

Janooo

Golden Member
Aug 22, 2005
1,067
13
81
When was the last time when NV came with smaller process before ATI/AMD?
If GF beats TSMC in 28nm then NV is going to have tough times ahead.
If TSMC is the first then still it's a good chance that AMD will be coming prior to NV.
I am not sure if Kepler can beat AMD's SI.
 

Lonyo

Lifer
Aug 10, 2002
21,938
6
81
FX5800 was on 0.13 while the 9700 stuck to 0.15
High end, the GTX280 was behind. GTX480 was behind.
8800GT (65nm shrink of the G80) was out a month before the 55nm HD38xx series.

5800 130nm vs 150nm 9800
6800: 130nm vs 130nm X800
7800: 90nm vs 90nm 1800
8800: 90nm vs 80nm 2900
G92: 65nm vs 55nm 3800
GTX280: 65nm vs 55nm 4800

The last time NV was first at the high end was the 5800 series (oops, and the 7800 series), although NV probably need the die shrink more than ATI does at this point, since they have a larger die so they cant do all that much more with 40nm.
 
Last edited:

Janooo

Golden Member
Aug 22, 2005
1,067
13
81
FX5800 was on 0.13 while the 9700 stuck to 0.15
High end, the GTX280 was behind. GTX480 was behind.
8800GT (65nm shrink of the G80) was out a month before the 55nm HD38xx series.

5800 130nm vs 150nm 9800
6800: 130nm vs 130nm X800
7800: 90nm vs 90nm 1900
8800: 90nm vs 80nm 2900
G92: 65nm vs 55nm 3800
GTX280: 65nm vs 55nm 4800

The last time NV was first at the high end was the 5800 series, although NV probably need the die shrink more than ATI does at this point, since they have a larger die so they cant do all that much more with 40nm.
So essentially 7800 was the last time NV came sooner than ATI/AMD on a given process. 5 years is a long time. If that's an indication then NV could be the second on the 28nm again.
 

Lonyo

Lifer
Aug 10, 2002
21,938
6
81
It's usually more due to timing than anything else.
With the GTX280, NV chose to stick with 65nm over 55nm, typically it's more a timing thing. One company usually comes out with a product before the other, and hence the timing of nodes and timing of product release dictates what process you use, not necessarily conscious decision.
The only time that isn't true is when a company holds back and sticks with an older process, which NV won't be doing (and neither will ATI).

If the number of transistors wasn't enough to turn this thing into a dinner plate sized bit of hardware, the fact that it's fabbed on a 65nm process definitely puts it over the top. Current CPUs are at 45nm and NVIDIA's major competitor in the GPU market, AMD, has been building 55nm graphics chips for over 7 months now. With so many transistors, choosing not to shrink their manufacturing process doesn't seem to make much sense to us. Smaller fab processes offer not only the potential for faster, cooler chips, but also significantly reduce the cost of the GPU itself. Because manufacturing costs are (after ramping production) on a per wafer basis, the more dies that can be packed onto a single waffer, the less each die costs. It is likely that NVIDIA didn't want to risk any possible delays arising from manufacturing process changes on this cycle, but that seems like a risk that would have been worth taking in this case.
http://www.anandtech.com/show/2549
 

Janooo

Golden Member
Aug 22, 2005
1,067
13
81
It's usually more due to timing than anything else.
With the GTX280, NV chose to stick with 65nm over 55nm, typically it's more a timing thing. One company usually comes out with a product before the other, and hence the timing of nodes and timing of product release dictates what process you use, not necessarily conscious decision.
The only time that isn't true is when a company holds back and sticks with an older process, which NV won't be doing (and neither will ATI).


http://www.anandtech.com/show/2549
There goes the question. Why did they do that?
In the last five years they seem to be hesitant or the transition is not that flawless.
Could it be linked to the design?
 

gorobei

Diamond Member
Jan 7, 2007
3,770
1,212
136
There goes the question. Why did they do that?
In the last five years they seem to be hesitant or the transition is not that flawless.
Could it be linked to the design?

more likely the process node. the smaller you get the more problems with materials and tolerances. 28 and 22nm will probably be as much of a crapshoot as 40nm, or else tsmc will have to delay them to make sure they are mature.
 

OCGuy

Lifer
Jul 12, 2000
27,224
36
91
When was the last time when NV came with smaller process before ATI/AMD?
If GF beats TSMC in 28nm then NV is going to have tough times ahead.
If TSMC is the first then still it's a good chance that AMD will be coming prior to NV.
I am not sure if Kepler can beat AMD's SI.

Are you sure? According to Charlie, nV will benefit just as much from a new GF node. Although he says it would be for Tegra.

But I don't count that site as a source, so it could all be BS anyway.
 

Janooo

Golden Member
Aug 22, 2005
1,067
13
81
more likely the process node. the smaller you get the more problems with materials and tolerances. 28 and 22nm will probably be as much of a crapshoot as 40nm, or else tsmc will have to delay them to make sure they are mature.
Good point, it seems AMD is the company that helps in the process maturation which means they could be releasing their 28nm parts before NV with TSMC or GF. The same stuff as the last fall.
 

Castiel

Golden Member
Dec 31, 2010
1,772
1
0
cef7pfquatgu.jpg


GeForce 500 series of products actively deployed at the same time, NVIDIA is also secretly force, and pay close attention to the next generation of new technology and new core. Latest news, NVIDIA has been very close to the 28nm process GPU core "Kepler, "the current film, this most crucial step to be completed.

NVIDIA 28nm chip foundry TSMC to continue to the old man, who has completed a number of new silicon chip technology networks work, but much more complex GPU graphics chip, easily several hundred million transistors, the natural need more time

Expected that the new "Kepler"family will be released in the fourth quarter, or this fall season, the first high-end desktop, or the landing area, and then gradually spread to the mainstream and mobile markets.

AMD's next generation of "Southern Islands " will be manufactured by TSMC is responsible, not to say that according to the prior GlobalFoundries, mainly AMD no resources to undertake R & D engineers, the two teams, but concrete progress is unknown. There was news that Radeon HD 7000 series will be available the second quarter of this year, but clearly too optimistic, according to common sense still have to wait until at least the end of the second quarter

http://vga.zol.com.cn/214/2144059.html
 

ShadowOfMyself

Diamond Member
Jun 22, 2006
4,227
2
0
Ah, yeah, the graph that lists Fermi as being a 2009 chip lol counting on that, except Kepler to be delayed to 2012
 

Wreckage

Banned
Jul 1, 2005
5,529
0
0
So in other words, I will have to disconnect from the internet for a few months at the end of this year. I don't need a new video card, I don't need a new video card, I don't need a new video card, I don't need a new video card, (I bet I will want one though).
 

Acanthus

Lifer
Aug 28, 2001
19,915
2
76
ostif.org
With TSMCs failure to get a die shrink out last year... I can't get too excited about 28nm until I hear they are actually mass producing ANYTHING on 28nm.

I expect delays.

It sucks because it is really hurting the march forward in graphics card performance (and smartphones for that matter).
 

Idontcare

Elite Member
Oct 10, 1999
21,110
59
91
I didn't see the other thread. My bad

I don't/didn't expect you to, but if it isn't materially different then I should merge it is all I was getting at. Doesn't mean the topic isn't worthy of continued discussion (hence I am not mentioning "lock" at all).

Moderator Idontcare
 

Ajay

Lifer
Jan 8, 2001
16,094
8,109
136
From the Beyond3d forums it looks like Kepler will be Fermi2, not a new architecture (have to wait till Maxwell for that). That said, nVidia will be able to pack allot a shader modules in @ 28mm with good thermals if TMSC's 28mm process is decent.