Personally I have to admit that I was a little reluctant to post news two days ago already as the source is Theo Valich (= ex-inquirer). However, what the heck ...
As you guys know NVIDIA's upcoming next-generation graphics processor will carry codenamed GT300 (GeForc Tesla) and is on a stampede course to be released later this year. (GT200 is GTX 200 series BTW).
Accoring to good old Theo, NVIDIA has already taped out some of the first engineering samples of the GPU, tagged under revision A1. This means, NVIDIA is likely already receiving it's first batches of the chip for a while.
It's pretty big news if it is the case as NVIDIA is then delaying the release of this product in order to go really big with it's launch. It will be the first GPU on the 40 nm silicon process for NVIDIA.
The GT300 comes in three line-ups, a consumer graphics, GeForce series, a high-performance computing product for Tesla, and G200GL for professional/enterprise graphics, Quadro series.
Also the rumor that the GPU will have 512 Shader processors, and that's a big step up from the current 240 Shader processors, is substanciated.
Next to that, NVIDIA supposedly is moving from SIMD to MIMD architecture. Long story short: SIMD always has the advantage in raw peak operations per second. After all, it mainly consists of as many adders, floating-point units, shaders.
The new GPU will have 16 clusters of 32 shaders. Though it is unknown how fast they will be clocked, rumors state that if they can be clocked at 2 GHz, we'd be nearing a cool 3 teraflops of compute performance (right now we are hovering around one teraflop) per GPU. GT300 would give direct hardware access [HAL] for CUDA 3.0, DirectX 11, OpenGL 3.1 and OpenCL.
Also it is now rumored that the GPUs will get a 512-bit wide GDDR5 memory interface to spew out around 256 GB/s of memory bandwidth.
There is no actual release date on GT300 but we do know it'll be somewhere this year, likely closer to Christmas and the Windows 7 release.