nVidia GT300's Fermi architecture unveiled: 512 cores, up to 6GB GDDR5

[Nemesis 1]

Well if AT article is correct . and Nv launches in feb . Thats impressive . I don't believe NV will get it right until the 4 tap out which would be may or june . If thats the case wait for the ATI 100. do in qt3 2010.

http://www.nordichardware.com/news,9997.html

http://www.nordichardware.com/news,9998.html

 

[ChaiBabbaChai]

...if they get the process to yield... http://www.semiaccurate.com/20...akes-fermi-boards-gtc/

 

[BFG10K]

Originally posted by: BenSkywalker

I have provided the financial reports to the six largest PC publishers in the world,

And? Where?s the proof they form the majority of the PC gaming market? Do you have a figure showing the total market so we can compare your six?s total to it?

Without it you?re simply speculating based on the six links you provided, six links which are an incomplete picture of the total market.

you can not provide a single factual number at all(neither can your sources as they flat out lie, they are a marketing group).

There?s a difference between marketing and flat-out lying. You?re basically saying they came along and said ?hmm, what figure should we make up this year? I know, we?ll make it 18% larger than last year!?

Sorry, if you think the figures are made up then you need to provide evidence of that. Showing six links and guessing they form 90% of the market is not evidence, it?s speculation.

You think Valve and Zenimax are larger then EA and Activision by a staggering amount?

I don?t know, why don?t you tell us? You?re claiming your figures are complete, so why don?t you tell us where those two rank? If you can?t tell us then your figures can?t be complete then, can they?

Again, guessing your six form 90% is not evidence, it?s speculation.

Talking about id is redundant, they are a subsidiary of Zenimax and not a private company. Another note about id, they aren't publishing through Activision anymore, the last game that they will publish before being solely Zenimax is published by EA.

That?s kind of my point.

No, it isn't in the slightest. They are "guessing" for marketing purposes, nothing more. Go ahead and check for their sources on their, heh, 'study'. You won't find it,

Not disclosing their sources doesn?t mean they?re wrong.

although you can easily check NPD or check with the SEC about the numbers I've linked.

NPD has already been covered why it?s inaccurate. Your own link stated they don?t include digital sales for heaven?s sake. Why do you keep repeating those results?

As for SEC, you?ve checked every PC developer/publisher in existence, have you? If you have, what?s the grand total you arrived at for the entire PC gaming industry, so we can compare it to the six?

You don?t even have Microsoft Game Studios for heaven?s sake. Any other serious omissions you?d care to mention?

They are a bunch of advertisers who lie throught their teeth, real numbers show them for what they are.

Again, the burden of proof is with you if you claim they?re outright lying. Posting six links and guessing they form 90% of the market is not evidence, it?s speculation. Until you provide evidence then there?s no point in discussing this with you further.

 

[MODEL3]

http://www.xbitlabs.com/news/v...g_Graphics_Cards.html#

At its GPU Technology Conference (GTC) last week Nvidia specifically noted that it paid a lot of attention on boosting double-precision floating point computational performance on its Fermi-G300 chip (about 750GFLOPs) and thus will be able to address new several new markets, e.g., high-performance computing. However, DP performance is rarely needed by average consumers on the desktop or laptop markets. Obviously, in order to create power efficient version of Fermi for notebooks or low-cost desktops, Nvidia will have to sacrifice some of its capabilities.

?We're not talking about other (chips) at this point in time but you can imagine that we can scale this part by having fewer than the 512 cores and by having these cores have fewer of the features, for example less double-precision,? said Mr. Dally, who did not explain how it is possible to reduce double-precision floating point performance without decreasing single-precision point speed, something which is needed by video games. In fact, Mr. Dally?s comment may imply that non-flagship Fermi derivatives will have not only be slower in terms of performance, but will be seriously different in terms of implementation.

EDIT*
Like i said, the value parts probably are not going to have the ECC features and these FP64 ratios.
Now we have confirmation.

But do i smell something more here?
Is the flagship model (Fermi, GT300) architecture much more modular in relation with the past architectures and in way NV can alter much more easily and cost effectively than the past the design?
(I mean having 2 designs, one for Tesla sector, one for classic Consumer sector)
What do you think?

 

[akugami]

Originally posted by: MODEL3
Like i said, the value parts probably are not going to have the ECC features and these FP64 ratios.
Now we have confirmation.

But do i smell something more here?
Is the flagship model (Fermi, GT300) architecture much more modular in relation with the past architectures and in way NV can alter much more easily and cost effectively than the past the design?
(I mean having 2 designs, one for Tesla sector, one for classic Consumer sector)
What do you think?

I was going to link to the same article and am wondering the same thing. At first, it sounded like nVidia was just putting out marketing speak and harvesting GPU's but more and more it sounds like the Fermi architecture is modular and they can cut out stream processors and other features to create smaller and cheaper GPU's for lower end video cards.

I wholeheartedly would agree with a modular design if they use it to cut costs and pass that down to users. For instance, 90% of the gamers don't give a crap about GPGPU and if cutting that out reduces the GPU size and makes it cheaper to produce and sell to the end users then I'm all for it. Likewise for someone looking to give their CPU's a boost by using GPGPU, they don't need high end gaming performance.

We've also seen how nVidia has had trouble scaling down the GT200 cores for lower end video cards as well as for the mobile sector. A more modular design should help in this regard as well.

 

[bfdd]

BenSkywalker, you hit a pet peeve of mine.... Activision does NOT own Blizzard. Vivendi bought controlling stock in Activision and merged Activision and Blizzard together. Neither own each other and Vivendi owns the new company ActivisionBlizzard.

 

[BenSkywalker]

And? Where?s the proof they form the majority of the PC gaming market?

Nothing publicly available, the numbers don't become publicly available until a couple years later. The reports run ~$3K per segment to look at.

Going to go with a slightly different approach since the main logical one isn't working. Let's take a look at PCGA's numbers shall we? 2007 numbers.

The summary of these findings
are that PC Gaming is a substantial
$54 Billion per year industry.

Gaming software (retail, digital
distribution, online subscription,
advertising and other business
models) generated $10.7 Billion
which according to most industry
studies would make PC game
software responsible for nearly
one third of every dollar generated
in the gaming software industry.

So according to them PCs sold $10.7Billion in software, and $43.3Billion in hardware. That's 214Million pieces of software at $50 a pop, or $200.93 spent on hardware for every $50 that was spent on software. This is using the numbers you are supporting. Furthermore, Asian/Australlian online sales excluding Japan are by far the largest segment in the world- they in fact dwarf US and European numbers combined(2.999Billion versus .776 and .86Billion). This is excluding Japan, convenient since there are no tracking bodies for those segments of the world outside of corporate financial reports(my numbers do list revenue from that region). One of the retailers they do track, and take the time to mention they track, is 7-11. I know they don't have them in your neck of the woods, but I can assure you Americans should get a chuckle out of that.

Another great indicator of their honesty, they quote North American and European revenue for all of the different platforms. As you have pointed out, along with myself posting it at the start, NPD only tracks a certain percentage of US retailers. The PCGA quotes sales data of the different platforms, it reports console sales for NA/Europe combined at quite a bit less then what NPD tracked in its limited capacity just for the US- link. In the US retail space that NPD tracks alone, the Wii did close to $3Billion more then what PCGA claims they did in total for both the NA and European markets.

Why is that?

These are the numbers you are choosing to support, so how on Earth is it that people are spending ~$200 in hardware for each game they are buying? Why is Europe and all online transactions making Billions of dollars vanish from the consoles?

The PCGA flat out lies, constantly. I have given you two links where you can start to gain a vague appreciation of the scope of their lies. They do not hesitate to misreport numbers by the billions per market to make themselves look better. These are the people you are putting your faith in.

 

[BenSkywalker]

Activision does NOT own Blizzard.

Welcome to the world of high finance, yes, they do-

Link. When two companies merge the company with the highest market cap at the time of the merger is the dominant party and hence the acquirer. Vivendi is a majority shareholder, but they are just that, a shareholder. In the financial world, Activision, without a doubt, acquired Blizzard. You may not care for the terminology, but it is accurate.

 

[cbn]

Originally posted by: Kakkoii
http://brightsideofnews.com/ne...c-up-to-6gb-gddr5.aspx


I know for sure I'm waiting to buy this baby.

"TSMC was in charge of manufacturing the three billion transistor mammoth, but it didn't stop there. Just like the G80 chip, nVidia GT300 packs six 64-bit memory controllers for a grand total of 384-bit, bringing back the odd memory capacity numbers. The memory controller is a GDDR5 native controller, which means it can take advantage of built-in ECC features inside the GDDR5 SDRAM memory and more importantly, GT300 can drive GDDR5 memory in the same manner as AMD can with its really good Radeon HD 5800 series. The additional two memory interfaces will have to wait until 28nm or 22nm full node shrinks, if we get to them with an essentially unchanged architecture. You can expect that the lower-end variants of GT300 architecture will pack less dense memory controller for more cost efficiency, especially on the memory side."


What do they mean by additional two memory interfaces?

 

[BenSkywalker]

What do they mean by additional two memory interfaces?

What they mean is that the chip is set to support 512bit memory interface but it isn't being used until future die shrinks based on what you quoted. 8x64=512bit, 6x64=384bit.

 

[coreyb]

definitely waiting for this card. I have a 285 right now and does the job nicely for every game I play. no reason to upgrade to a dx11 card yet.