| Aielyn said: "The answer comes from a mixture of known and unknown variables." |
OK look here is the deal. The most powerful R700 GPU ever made on 40nm node that approximates the 150mm2 die size of the Wii U's GPU is RV740 (HD4770).
That GPU has 960Gflops of processing power, 750mhz GPU clock and 51.2GB/sec memory bandwidth and on its own used 80W of power. I believe the power block for the entire Wii U console is just 75W. Even if we assume that the Wii U has an HD4770 with all of its shaders, textures and ROPs intact (640:32:16), it still has a 27% lower GPU clock and just 12.8GB/sec memory bandwidth, or 75% less than HD4770's. That means in the best case, it would have just 50% of the power fo the HD4770 because dropping memory bandwidth from 51GB/sec to 12.8GB/sec is suicidal for GPUs. R300 Radeon 9700Pro GPU from 2002 had almost 20GB/sec memory bandwidth.
Half of HD4770's performance is not a next generation console and this is the absolutely best case for Wii U's GPU. The rest of the specs are just minor details that do not change the conclusion that the GPU has to be less powerful than HD4770, a lot less. Also, it's pure physics. You cannot make an R700 GPU that used 40-50W of power that's as powerful as an 80W one. By extension of Wii U's maximum power consumption, no R700 GPU inside of it made on 40nm process can ever even touch HD4770.
