Well I think we are nearing the speed limit here. I mean using the current x86 base we can only achieve so much. By phasing out x86 instructions and replacing them with an instruction set that can take better advantage of current processes I think we can see some real gains.
I really have to wonder what could be done if we completely changed the way registers were addressed?
The architectural advantages (and current deficencies) are way too much for a forum post or even a short article but do know AMD is thinking about making a move (or at least brodening theie instruction set to include ARM) and a I heard one or two things pertaining to Intel as well.
NV'S TEGRA3 looks good but i want to see it vs the new TI chip due out sometime next year. according to some numbers the power-VR video processor is suppose to be twice as powerfull then the one in tegra 3.
This site uses cookies to help personalise content, tailor your experience and to keep you logged in if you register.
By continuing to use this site, you are consenting to our use of cookies.