Seamless offloading of web app computations from mobile device to edge clouds

Wifi is not a bigger drain than cpu when you're playing CoD on your mobile. Your smart phone now has multiple cpu's and would have a GPU. When you're doing something intensive like CoD it will fire up that GPU and suck up power.

Wifi power used will vary with a few factors, how far you away from the station, how many other wifi signals, etc.

In addition a big bright screen uses a lot more power than wifi. But that would be used either way so not part of the comparison.

> I can play call of duty on my mobile with graphics better than xbox 360

I noticed graphics on smartphones are pretty damn great (I don't play mobile games) these days.

So, the Snapdragon 855+ has a GPU that does 1030 GFLOPS, which is very close to the XBox One (1300 GFLOPS), at a fraction of the power usage. Amazing!

According to them they can support 2x the objects with acceptable perf on the server compared to the client.

It appears you intuition is wrong in this case.

Thermals too, keeps your phone from getting hot?

It seems to me they are suggesting adding more powerful processors to edge devices such as routers if this turns out to be a useful technique.

Those other devices aren't running on battery power, which could be a significant factor.

Also, there are many less powerful phones out there which may see greater benefit.

And the "edge device" presumably doesn't have to be the absolute edge, perhaps another part of the edge infrastructure (in my home this could be the server that sits directly behind the wireless AP topologically speaking.

And some APs do have significant computing power, they may not be taking about the cheapest router in existence provided by a penny-pinching ISP.

> processing power is rarely a major limiting factor.

Autonomy is. I could see compute clouds like this being offered at the OS level rather than browser, along with a claim for multiple days charge cycles, or slashed device weight/thickness. However, it would likely only work well for common and predictable computing tasks. And it would further separate users from hardware.

Deep learning can be used for many interesting things, like computer vision applications.

Even visual odometry and SLAM, necessary for AR, can be solved through DL.

I, for one, can't wait to have something in a raspberry Pi form and power factor that can run a big deep CNN at 30fps.

That opens the way to a lot of neat gadgets that can be summarized as "computers that can see":

https://www.ben-evans.com/benedictevans/2019/7/19/computers-...

Phones come packed with quite a few sensors and communication capabilities (multiple cameras with parallax and varied optical parameters, microphones, accelerometers, GPS, depth sensors sometimes, etc.). There has been a lot of focus over the years to do useful things with commodity sensors since they're so pervasive.

There are a lot of very sophisticated and computationally intensive algorithms out there that can use specific sensors or combinations of these sensors to infer other useful things.

Now take into account you may have multiple phones collecting data distributed in some spatiotemporal fashion, you can really get fancy with some aspects and increase accuracy of some of these algorithms.

Instead of hand waving, one novel example I've worked with is using video feeds from streams or rivers to perform LSPIV (large scale particle velocemetry) analysis to estimate instantaneous discharge rates of rivers. There are at least 6-7 other use cases for projects I've been involved with that could leverage these resources in interesting ways--some use phones, some use other sensors.

Some phones are strong. Most are cheap and crappy.

It really is a pendulum isn’t it? Is it driven by cost dynamics? Computers become expensive and incentivize sharing (mainframes, cloud, sevens, ...). Prices come down — either because workload become more complex or hardware becomes cheaper — and the incentives for exclusive ownership come back.