but we're talking about handheld mode (the mode where power draw would even matter). near-docked clocks at handheld more for the same power draw or above-handheld mode clocks for slightly better performance and better battery life. or option 3: same clocks, better battery, no active cooling for handheld mode
I'm just gonna quote my post again because you always distort the meaning of anything that isn't quoted each passing post. What you said is basically like saying "If I measure the Switch playing Breath of the Wild for one minute and idling in the menu for another minute and make an average, it averages 8W". It's meaningless. Not long ago you were saying the 7.5W figure for the Jetson Tx2 was for the whole dev board, then you said it never reached its specified TDP, now you bring up the average of that measurement where only a fraction is even compute-heavy.
You can clearly see a stretch of time where a GPU-only load draws close to 8W, with the peaks to 8.2W you mention. That is, without putting load on the CPU, and without moving heavy assets around memory, since it's a synthetic matrix sum and multiplication test. I think it's safe to assume that, under the load of a demanding game, the Jetson Tx2 would draw above that "close to 8W" figure at Max-Q clocks.https://www.phoronix.com/scan.php?page=article&item=jetson-tegra-x2&num=3
That's power consumption during bus read and write, texture read and GPU compute test, for the Jetson Tx2 in Max-Q mode. (The closest we have in the Linux space to a 16nm Tegra X1, very close indeed) The module draws about 7.8W on average during the GPU compute test, even when the CPU is basically idling, as shown by it only consuming 0.4W at the same time. Those 7.8W are also what the combination of SoC, DRAM and wireless controllers in the Switch must use under full gaming load. So go figure, to me it looks like it wouldn't be much of an improvement at all.
It's not perfect but this is what I've been picturing when I say clamshell:
It's not perfect but this is what I've been picturing when I say clamshell:
They'd have to change some of the game physics and gameplay for this as most interactions and enemies are optimized for 30fps. It's not like Fortnite or Doom. I think it would be more hassle for them in the long run, and its Nintendo of all people who are slow to the optimization party. At best, I could see Nintendo increase the resolution to a solid 1080p and stable 30fps framerate all around, and maybe reduce pop ups/increase draw distance. I'd be more than okay with that. But even then sadly I think expecting Nintendo will do this is just keeping my hopes up too high. I'd replay botw again for sure if they did this.
I doubt it's Tegra X2, simply because there is no code for Tegra X2 in the firmware but there is code for Tegra Mariko, whatever that ends up being.
Yeah they might as well put a cheap touchscreen in there for 3DS virtual console or games with optional second screen functionality.
I like this. I wonder how flexible their ability to feed out of that usb-c port is. I'd love to have basically this as a setup with a battery and screen in the bottom half. That's amazingly sweet.
What's the highest clock settings that can be done with the mod? Is the max 768 mHZ/393GFLOPs?
It should be the same or cooler than the temp of the Switch when simply docked, since the dock doesn't help cool the system.
Obviously if a game runs at maximum resolution of the screen (720p) and at the maximum framerate selected on the game (60 or 30fps), then it will not be a improvement.
Just wondering if you or anybody has tested the switch at max clocks. Would be interesting to see performance in framerate and resolution, even if its a 30% increase above Switch/Nvidia Shield's bottleneck at 393 GFLOPS/768 MHz, as well as temperature and watt usage...
That would be a great video!
I see. That battery life is low, but it is interesting to know that the Switch is sufficient enough to run games at the docked clock speed without visible short-term damage
Try sending it for repair to Nintendo, maybe it's not something major.
We don't. All rumors for now. I think Q4 2019-Q1 2020 is a fairly safe bet.
