se7en7thre3 said:
Obviously being a mobile design, would knock it down in performance but if you had 2 of these suckers for the NX home console version, then it may match/exceed XB1. A downclocked version of the x1 chip in the NX handheld, and 2 of full versions (if not one, newer designed chip) in the home version would be ideal. If Nintendo uses the same chips in the handheld and other iterations, they will get a better manufacturing deal. The key is to get the HH version powerful enough for this approach to be feasible. |
Nvidia Tegra X1 : 256:16:16 (1000mhz core) 16 GPix/s , 16 GTex/s , 25.6 GB/s Memory bandwidth, 512 GFLOPS (FP32). 3GB of ram?
Xbox One : 768:16:48 (853mhz core) 13.6 GPix/s, 40.9 GTex/s , 68 GB/s DDR3 (+eSRAM ~109 GB/s), 1.31 TFLOPS (FP32). 8GB of ram.
~2.5x GTex/S , I dont think how many Gflops of compute a GPU has is all there is too it, GTex/S is something that matters too.
(177 GB/s vs 25.6 GB/s) => x 6.9 memory bandwidth
about x 2.56 times as much compute power inside the xbox one GPU:
http://www.anandtech.com/show/9289/the-nvidia-shield-android-tv-review/9
Power consumption:
Xbox One = ~100watts
Tegra X1 = ~19.4watts
People make it sound like the Tegra X1 is basically a Xbox one.... thats simply not true.