The upcoming launching of the new chip for high-performance 3D graphics accelerators from NVIDIA, aka NV30, is rumored more and more with new details cropping up now and then. The discussion of the potential NV30 features, which has been resumed stimulated by the SIGGRAPH Conference, revealed some new specs details. Here is what beyond3d.com guys write about NV30:

• .13u process
• AGP 8x support
• DDR2 Memory support
• Vertex Shaders beyond DirectX9 VS2.0 - supporting up to 1024 static instructions, 65536 instructions executed in loops, branches, and subroutines
• Pixel Shader beyond DirectX9 PS2.0 - with up to 1024 instructions
• Support for OpenGL and DirectX HLSL via Cg
• OpenGL Extensions supporting long Pixel Shader & Vertex Shader programs
• High Precision (64- and 128-bit Floating Point color)
• New focus on computational efficiency rather than memory efficiency
• Advanced programmability and high-level shading language support
• Expected this fall along with DirectX9

In fact, there are two conclusions we can draw out of these features. Firstly, NV30 will be a highly configurable and programmable chip. Cg language seems to have been launched exactly for NV30. The developers willing to use the features of NV30 to the full extent will have to resort to Cg.

Secondly, NVIDIA refused to continue working on technologies unloading the memory bus. That is why the developers will introduce DDR-II support. This will allow the company to equip its graphics cards on NV30 with much faster memory chips, than you can imagine today. In particular, according to the preliminary standard approved by JEDEC, the DDR-II graphics memory will be able to run at 800MHz and 1000MHz.

However, there is not so much waiting left before we will be able to see what the actual NV30 is like. From what we hear, the new NVIDIA solution is due in October this year.