Just wondering what the difference is if any between Pixelpipelines and Stream processors. I see that the pixelpipelines only go up to 24 units on the most expensive cards but stream processors or processing units can go up as high as 800 units at the moment.
So i guess what im asking is what are they. Should one method be chosen over the other for gaming. And lastly is it as key of a stat or even a bigger stat to look at as far as performance as memory and bits are.
Thank you for any help that you can give me as im a little confused.
Back in the olden days (meaning just over a year ago) we had different shaders depending on what needed work at the time. There were pixel shaders and Vertex shaders, which got used depending on what needed to be worked on at the time. This is bad, because if you needed at lot of vertex shading power, your pixel shaders will be sitting around doing nothing. Enter the stream processor, or Unified Shader Architecture. (USA)
With stream processors, it doesn't matter what you are "shading". Pixel, vertices, it doesn't make a difference. The operation is similar enough that the stream processor can do whatever.
In the end, it doesn't matter how you do it. USA is pretty much "required" for DX10. (its not really, but heavily encouraged.) Separate pipelines have gone the way of the dodo, so embrace the new technology. That doesn't mean that pixel pipeline cards are weaker then SP cards, it depends on how many SP/pipelines each card has.