GM200 confirmed for 2014 launch

[SoulsCollective]

via VideoCardz:

Paper titled ?Suitability of NVIDIA GPUs for SKA1-Low? has revealed some NVIDIA?s plans for the upcoming Maxwell and Pascal architecture.

This document would probably mean nothing, if one of the authors wasn?t working for NVIDIA itself. Mike Clark, who is Compute DevTech Engineer at NVIDIA, has contributed to this research, so it would be rather shocking if the data he provided would be bogus.
...
The most interesting part is where GM200 pops up. According to this document, the Big Maxwell would arrive yet in this year. This GPU would be twice as power efficient as GK110. Of course this doesn?t mean it?s twice as fast in games.
It means that with the same power consumption, it will deliver 208% of GK110′s synthetic performance. This is probably where we reach 28nm node limits. GM200 silicons are still in development stage. It will take only few months for GM204 to launch since we first saw it at Zauba. GM200 appeared just 2 months later. Late 2014 launch seems plausible.

[Lanek]

We was discussing the paper in question some days ago..

You can download the paper directly from Cornell university site and read it for do your own opinion : ( i put the link directly but it is in the videocardz article too. ) http://arxiv.org/abs/1407.4698

[zalbard]

The paper proves nothing... It's just what a single researcher thinks.

[Iconyu]

I don't see Nvidia releasing the GM204 and GM200 within 3 months of each other as consumer parts, if the dates are true we're simply looking at a new Tesla before the end of the year.

[Russian]

Can someone please, for the love of god, explain to me why we're talking about GM2XX when we only have seen one GM1XX chip? I don't think any of this is right. It has to be GM1XX in 2014 and GM2XX in 2015 and GP1XX in late 2015/Early 2016....

[Chrono Detector]

Meh... Videocardz, not sure if they are very reliable with their sources, but if that is true than that's good news.

[Lanek]

The paper proves nothing... It's just what a single researcher thinks.

Effectively, hence why i have suggest peoples to read it before jump on conclusion. Everything look more speculative on the GPU part.. The point was more to make a mathematic method for calculate the ressource of the Low-SKA1 station ( astronomy station driving 100+ Antenna ).

[SoulsCollective]

The fact that an nVidia employee /wrote/ the thing suggests that they know what they're talking about. Now, given that it's talking about compute performance, this in no way means that we'll see an 880 Ti launching this year, but I would not be surprised to see GM210 Quadro cards come out.

[Lanek]

The fact that an nVidia employee /wrote/ the thing suggests that they know what they're talking about. Now, given that it's talking about compute performance, this in no way means that we'll see an 880 Ti launching this year, but I would not be surprised to see GM210 Quadro cards come out.

You dont really know what he's know, and what he was knowing when they have start to write this paper. It have been first published in July, but they was maybe work on it allready since 4-5 months ( recover the actual data, create a method of calculation, make valid them etc etc ).
Then we dont really know what is the participation of him on the research.

Ofc, looking at the first GK110 Tesla, we remember first chips have been send in summer 2012 to some specific supercomputer project who was in phase of building the system. Then nearly 6month after this, we was seeing the official release of this Tesla, followed by Titan.

But all of this can be based on previous internal roadmap.... then the system is really specific to this appliccation system, driving Antenna monitoring station is not really comparable with other HPC syntethic benchmark.

[Tenknics]

Another non/news thread about Maxwell/Nvidia rumors? DUDE, this is getting realllllly old. Can you not keep these kind of posts in the rumor/speculation threads that already exist?

Why is there such a compellling need to start a new topic everytime you read something about Maxwell? do you need attention? This is getting so old and frustrating.

This is not news! These sources you use are not credible! This does not belong here!

[LordEC911]

Ofc, looking at the first GK110 Tesla, we remember first chips have been send in summer 2012 to some specific supercomputer project who was in phase of building the system. Then nearly 6month after this, we was seeing the official release of this Tesla, followed by Titan.

They shipped their first batch of GK110 to Oak Ridge in September '12.

[Lanek]

They shipped their first batch of GK110 to Oak Ridge in September '12.

YOu right, i was not completely remember the date.

Anyway, nobody ask how a guy from a known forum of hardware ( in Deutshland it seems ) have suddenly got the idea to visit the Cornell University site and find an obscure publication about future ressources calculation of an obscure astronomy station system ( the SKA1-LOW ) and suddenly find information about Maxwell inside it ?

[astrallite]

Doubtful unless its a December 31, 2014 launch, lol...

I can't remember Nvidia ever replacing the fastest single card within 3 months of its launch with a new fastest card though.

[Lanek]

Doubtful unless its a December 31, 2014 launch, lol...

I can't remember Nvidia ever replacing the fastest single card within 3 months of its launch with a new fastest card though.

I dont think too, but the other way around, Nvidia could launch it as Tesla part...

A consumer part could come way later in end of Q1 2015.. Even marketing wise, this could benefit them on the consumer market. Whatever is the performance of the GTX880, they have something who will " come later ".. bigger, faster etc ...

[SoulsCollective]

WCCFTech:

Here is the shipping entry from Zauba.com that confirms that the GM200 aka the Titan II GPU are now exiting the internal testing state...Before the first ‘CS’ entry which appeared on 12th August 2014, there had only been entries using the nomenclature ‘INT’ which means the core is meant for Internal Testing. Because green follows the same pattern we knew that once the INT chips were spotted the CS chips wouldn’t be that far behind and it looks like our waiting has paid off because the chips with the name GM200-CS0-A1 and GM200-CS1-A1 have been spotted, indicating that the GPU is now more or less ready.

I'm not at all familiar with 'Zauba', so no idea how reliable this is.

[Mats]

The fact that an nVidia employee /wrote/ the thing suggests that they know what they're talking about.

Just because they know what they're talking about, it doesn't meant that they're telling the truth.
There have been a bunch of bad predictions through the years about upcoming hardware.

In this case, I have no doubts that the final product will most likely be a success tho.

[LordEC911]

"GM204" is "it" for Nvidia's highend until 2H '15.

[Andrew LB]

Doubtful unless its a December 31, 2014 launch, lol...

I can't remember Nvidia ever replacing the fastest single card within 3 months of its launch with a new fastest card though.

Think about it. Why would nVidia be bringing out new cards so soon? Simple. To cut manufacturing costs by transitioning to 20nm. You can fit a lot more 20nm chips on a wafer than you can 28nm. Also, TSMC's 20nm is in full production on two separate fabs, with a third on the way. And even then, they're completely booked through the end of the year, and one thing's for sure... it's NOT just apple buying all those wafers. Also, Apple recently took delivery of 20nm chips used in the upcoming iphone 6.
http://wccftech.com/tsmc-warns-capac...ing-3-billion/

[LordEC911]

Think about it. Why would nVidia be bringing out new cards so soon? Simple. To cut manufacturing costs by transitioning to 20nm. You can fit a lot more 20nm chips on a wafer than you can 28nm. Also, TSMC's 20nm is in full production on two separate fabs, with a third on the way. And even then, they're completely booked through the end of the year, and one thing's for sure... it's NOT just apple buying all those wafers. Also, Apple recently took delivery of 20nm chips used in the upcoming iphone 6.
http://wccftech.com/tsmc-warns-capac...ing-3-billion/

I haven't heard anything about a 20nm Nvidia GPU and I don't know why they would... especially when 16nm FinFet is right around the corner.