What an amazing idea...
Sensible, non-wasteful both financially and environmentally.
It's doomed. And the last humans will curse us all as they drown in a poisoning sea of old broken mobile phones.
A Finnish group of phone developers, hoping to get the world interested in modular smartphones, has proposed a nifty idea for re-using their phone motherboards: turn them into clusters. The Linux-based Puzzlephone project wants to extend the life of smartphones by making more of the phone replaceable, on the premise that most …
this makes about as much sense as people who used to want to collect a bunch of old computers to build a beowulf cluster or something. It's a cute idea, but beyond being cute it is quite pointless. (same goes for folks that like building clusters out of Rasberry Pis, cute for about 30 seconds).
You mean its pointless like how car manufacturers realised that instead of using a single cylinder, they could cluster them in a V, W, straight, or horizontally opposed configuration, and get far more life and performance out of said cluster, than they could from the single piston?
We need a car analogy icon.
You mean its pointless like how car manufacturers realised that instead of using a single cylinder, they could cluster them in a V, W, straight, or horizontally opposed configuration, and get far more life and performance out of said cluster, than they could from the single piston?
Or like how I travel to my destination twice as quickly when I drive two cars at once?
Because more of something is automatically better.
We need a car analogy icon.
Clearly, what we need is a stupid analogy icon.
If you dont believe that more of something is automatically better, what system, may I ask, did you write that reply on?
Was it a system with a multi-core CPU? Did it have more than 1gb of RAM? More than 1GB of hard drive space? Did it, perchance, have a multi-monitor display?
The stupid analogy is yours, good sir.
Was it a system with [blah blah blah rubbish]?
It's a system that was purchased by my employer; I didn't choose the specs. I still have computers in use with single-core CPUs and less than 1GB RAM. I don't think I have any with drives smaller than 1GB, simply because such drives typically died a while back and replacements are all larger.
And no, no multiple monitors. As I've noted in other posts, I had a multiheaded system back around 1990. I've not seen any need to return to that configuration.
If you dont believe that more of something is automatically better...
Then you're probably capable of critical thought. And indeed I am. You should try it sometime. (Also, may I suggest learning how to use the apostrophe?)
First of all we are far from a common hardware platform which would be needed for this. Our mobiles currently are just different enough to prevent you from installing a common operating system without having to port it for every single phone. Even large projects like Cyanogenmod can only do this for a small part of the market.
Second, mobile processors are rather slow. Combine that with the overhead, and you'll most likely end up with something that's much more expensive than just getting a computer that's much faster than the equipment to connect the mobiles. However this might work if you use already existing hardware. Your "rack" could consist of just some charger ports and you could do communications via Wifi.
Interconnect bandwidth would be pretty bad.
The real issue, though, is that intact phones are going to want to power up the display, radios,1 etc. So the performance/power ratio will suck, relative to the just-the-motherboards design being mooted here. You could run specialized software on the phones so you aren't powering any peripherals you don't need, but if you're going to that trouble, you've already incurred a significant fraction of the cost of dismantling the phone - particularly if the phone's designed to make that dismantling easy.
And connecting power to all the phones is going to require some rack-like arrangement anyway, unless you want a nasty tangle of cables.2 So what's the point?
Mind you, it sounds so much like a plot point from a contemporary TV drama that I bet some writer's already put it in somewhere. "Oh no! [Scary hacker team name] has taken over all the phones in Manhattan and networked them with Bluetooth to create a super computer cluster that they're using to crack the President's daughter's password!"
1You can't just put them in airplane mode because you're using Bluetooth.
2Or you're anticipating using wireless exclusively. That seems unlikely, and I don't know how efficient it is.
5 years is about the useful lifetime of most supercomputers. And the big cost is usually running a supercomputer, the ongoing power and cooling is easily 5+€/£/$ million a year. Throw in a 6+ support staff for 24/7/52 operation and the hardware is the cheapest part.
In reality, this is just advertising .
Personally, I've had to change phones for three reasons:
1. Spent a day crawling over concrete floors with phone in front pocket (don't ask). Screen was toast. Stupidity, in other words.
2. Dropped Sprint and switched to AT&T for GSM rather than an also-ran protocol that's useless in most places outside the US. Had to get a GSM phone, obviously.
3. After four years, screen on that phone had acquired too many bad pixels. Battery life was around four hours, but if the screen had still been good, I would simply have replaced the battery. Of course all my phones have had replaceable batteries.
I've never wanted to replace a phone because the motherboard was ... what ... too slow? Geez, here I am stuck with the equivalent of an '80s graphical workstation in my pocket. Life is hard.
If the goal is to build a mobile robotics platform, 120 S8 boards in a rack might be fine and draw a tenth of the power of an array of graphics cards, TPUs etc.
Also there are a lot around as users upgrade all the time, perhaps have an app that secure erases the phone once the data has been verifiably transferred to the new phone (independent of screen damage etc) so the board can be sent by post without issues.
The latest phones have AI inference chips on them already but the goal here is sheer processing power.
(wonder if I can get my hands on a couple of hundred Note 7 boards anywhere?)