Did LCGeek mention anything about power drain with the Tegra running at full tilt?
Nvidia did tell us that Pascal Tegra is 60% reduction in power consumption for same performance over Maxwell Tegra, or 40% greater performance at the same power consumption.
-X1 SoC is 10watts and was passively cooled in the Nvidia Shield TV AFAIK.
-NS is actively cooled at least when docked.
-Nate told us it is Pascal, and Nvidia seemly has backed it up but it is worded loosely and I think that is because Volta comes out next year and might make Pascal look dated.
-Laura Dale told us that the dock upclocks the switch for higher performance and that on the go, you have a 3 hour max battery life.
-N3DSXL consumes ~4 watts and has a 5 hour max battery life with a 1750mAh battery.
-Vita consumes ~7 watts and has a ~4 hour max battery life with a 2150mAh battery.
Assuming the above, the device is likely drawing ~7-8 watts on the go depending on the battery size. It's performance would be ~600gflops on the go and be passively cooled.
When docked, the full tilt of the chip would be anywhere from 1.4ghz to 1.6ghz, putting it at 717gflops to 819gflops. Since not all flops are equal and 16fp could be important with this design, the comparison to XB1 shifts the numbers like this: effective 1.144tflops to 1.3tflops and without 16fp you'd be looking around 1tflop. On the go, that 600gflops would effectively be just under 1tflop when compared to XB1's architecture or just over 800gflops with no 16fp usage.
Take the above as a loose estimation about what is possible given what we know, Nintendo could be using a smaller battery, or have much faster CPUs than found in the X1, but I doubt those things. Emily in April said that the Switch was going to be closer to the XB1 than PS4 and even that was a tiny bit of a stretch. When looking at the numbers above, I think that is the picture it lays out, these are similar performance to XB1 but noticeably below it.