haxxiy on 23 September 2015
I remember you! The guy who was all like "Tegra X1 is 1 TFLOPS" and then proceeded to be disproved because that was half-precision floating points. And that was without even factoring in power throttling and the way GPU makers like Imagination Technologies lag way behind AMD and Nvidia when comparing FLOPS to FLOPS.
Even at 16nm it is doubtful we can squeeze more than, say, 64 GFLOPS (AMD FLOPS) per Watt, and in average an iPad is designed to dissipate about 2.5-3 Watts (given their battery capacity and battery life). So no again.
But try once more when we get to 10 nm or so, by all means. One day we'll get there.
