Quote:
Originally Posted by paroh
The processor are so much small in size but the size of the graphic card are are so much bigger in size as compare to the processor.
I think now is the time that graphic card work can be taken up by the processor it self.
For example i7 contains 4-6 cores. now four cores for processing and other two cores for graphic works.
I think this could be possible if companies like intel or amd works with the graphic card manufacture like xfx or ati to build a such a processor.
|
It might be possible in the future, but as of now the GPU is better equipped to do graphic rendering. The CPU is the central brain of the computer, and is managing everything. The CPU infact send the 3D data to the GPU. When it does this, it offloads a major task to the GPU.
The GPU is good at this, because the core is designed in a different manner. It is not a 'manager'. It is a HUGE calculation engine, which is extremelly data hungry ---> power hungery. Cause...
Basically a lot of mathematical calculations are done to convert from one format to another. This can be off loaded to the GPU core. The GPU core is efficient at doing this, cause it can calculate to really small decimal places. Light vertex numbers are done using this. So the GPU is really good at number calculations.
They have the architecture of 'stream processors'. This is the SIMD (single instruction multiple data) logic that the GPU uses. So multiple data points are applied a single instruction set in one go, very quickly. And these are all put in a pipeline -- which is the pixel pipeline. So imagine a pipe full of data which needs to have floating point calculations done on it. The data comes out of the pipe to the GPU and the calculation is done. Now there are multiple pipes like this which feed data to the GPU, which does the stream processing. GPUs are rated with the number of pixel pipelines available to the GPU.
This enables the GPU to quickly and efficiently process data. Perfect for 3D rendering or format transformation.
It does all this, renders the 3D polygon image, and renders them on the screen.
So if you have 1400 x 900 resolution monitor and are getting 60FPS, the GPU is actually rendering a screen full of 1260000 pixels x 60 times a second.
To get a good idea of this, down load the Unigine benchmark took from
here. Run the demo as per your Dx version and OS, and press F2 to see the Polygon rendering system. You will immediately see the task of the GPU.
processor size vs graphic card size
Linear Mode
