The other thread is spreading out in many different directions so I thought that I would pose this question in a seperate thread.
While reading another site about peoples thoughts on Fusion there was a question whether the cpgpu could support a seperate gpu. I dont think that this would be a problem (but then I am no expert by any means). Now I guess this becomes what would then be done with the "spare" parts of the cpgpu once you have a high end board in the same system. Could the gpu part of the core then be used as a physics processor, stream processor, geometry engine...?
What kind of rewriting to the driver would be necessary and would it be worth it?
Will AMD still produce discreet high end cpu solutions just for the high end market where they would then become niche products like certain current high end graphics cards?
While reading another site about peoples thoughts on Fusion there was a question whether the cpgpu could support a seperate gpu. I dont think that this would be a problem (but then I am no expert by any means). Now I guess this becomes what would then be done with the "spare" parts of the cpgpu once you have a high end board in the same system. Could the gpu part of the core then be used as a physics processor, stream processor, geometry engine...?
What kind of rewriting to the driver would be necessary and would it be worth it?
Will AMD still produce discreet high end cpu solutions just for the high end market where they would then become niche products like certain current high end graphics cards?