I suppose that it is possible as you said, because the r580 had 3x the pipelines the r520 had. But the way I see it, the rv670 was a stepping stone towards getting the process right for fusion, and then the rv770 will bring some big performance to make a profit, and then the r8xx whatever fusion is will be a perfected form of the r600 design, whatever that may be. So I predict a very noticeable increase, but a 150% increase in shaders? With how little the dize size is changing, I find that hard to believe
Come to think about it, the r600 could be looked at like the r520. It lost to nvidia's high end, but isn't wasn't made to win, it was made more so as a test to pave the way for the next big bomb they drop on the market. Although not the best marketing strategy, when funding is low, that's what you have to do, and I actually wouldn't be surprised if the rv770 wows us more than we predict




Bookmarks