Nvidia's performance 40nm card is GT214

[explicit4u]

Nvidia looks very promising with the new 40nm tech.

Another 40nm


Nvidia is working on many 40nm chips at the same time and GT214 is the performance card that will come a bit later this year. According to our sources, this card should launch at some point in Q3, roughly a year after the 65nm GT200 was launched.

Traditionally, Nvidia launches its entry level and mainstream with the new technology and once the manufacturing process at 40nm is mature enough, they launch their performance cards. Nvidia sales really depend on the performance cards.

The new card is codenamed GT214 and it should come with a TDP of around 100W, but we don’t know much more than that. This is the last performance card from Nvidia to support DirectX 10 only, and the new card codenamed GT300 will have the full DirectX 11 support but it will came in the last months of 2009.

 

[mame]

I thought DX 11 was just a software update

 

[Lightnix]

I thought DX 11 was just a software update

Software that accesses hardware functionality, of course!

 

[HummuH1]

Not much to get excited about in that article.

 

[Helios1234]

Strange shouldn't the GT212 be the performance card like the g92 with the GT214 being the mainstream card like the g94?

 

[explicit4u]

Direct X 11 will not show any real improvement i think, more of a tweak, or 10.2

 

[Stonedofmoo]

Why is it DX10 only. That makes no sense at all assuming it will be sold for a year at least.

 

[Helios1234]

Direct X 11 will not show any real improvement i think, more of a tweak, or 10.2

This might change your mind: http://www.anandtech.com/video/showdoc.aspx?i=3507

Basically dx11 will make dx10 look like a mere stop gap.

 

[Lightnix]

Direct X 11 will not show any real improvement i think, more of a tweak, or 10.2

Hm, let's see here:

Shader Model 5.0: Supports object oriented shader code (makes developing much more complex shader code a lot more feasible)
DX11 compute shader: general purpose GPU programming language with tight integration to DirectX 11 (like OpenCL integrates well with OpenGL).
Hardware tessellation with programmable inputs and outputs: for all intents and purposes this is a hardware subdivide with programmable input and output, from what I've heard at least, means much smoother models with much more detail.

 

[Helios1234]

Hm, let's see here:

Shader Model 5.0: Supports object oriented shader code (makes developing much more complex shader code a lot more feasible)
DX11 compute shader: general purpose GPU programming language with tight integration to DirectX 11 (like OpenCL integrates well with OpenGL).
Hardware tessellation with programmable inputs and outputs: for all intents and purposes this is a hardware subdivide with programmable input and output, from what I've heard at least, means much smoother models with much more detail.

Yeah anandtech has a great article on it and it looks promising. Link 2 post above.