Modular Smartphones Could Be Reused As Computer Clusters

itwbennett writes The promise of modular smartphones like Google's Project Ara is that buyers will be able to upgrade components at will — and now Finnish company Circular Devices has come up with a use for discarded computing modules, which they're calling Puzzlecluster. Drawings of the Puzzlecluster architecture show a chassis with slots for the reused modules, which can then be interconnected with others to create the cluster. Just one unit could also be used as a desktop computer."

Comments

Oh please, you act as if they're computers

[WillAffleckUW]

Next thing you know, you'll tell me that the modern smartphone has more processing power and data storage than all the spacecraft we've sent to other planets combined, and all the computers we built up to the year 2000.

pointless

[itzly]

Clusters of underpowered processors are not nearly as useful as a single powerful processor.

I suppose...

[fuzzyfuzzyfungus]

Assuming that the obsolete compute modules are of standard size/pinout (or, more likely, that compute chassis are only produced for phones that ship in sufficiently massive volume to assure a supply of board-donors), this scheme would work; but I have to imagine that a phone SoC would make a pretty dreadful compute node: Aside from being a bit feeble, there would be no reason for the interconnect to be anything but abysmal. For efficiency's sake, SoCs tightly integrate all the parts that need to chat at high speed with one another(along with whatever else fits, just to save board space), and only such interfaces as are absolutely necessary are brought out of the package, much less broken out on the board in some sort of civilized connector. In terms of dedicated interfaces you'll have some dubiously appropriate wireless stuff, a USB slave or host/slave interface, and a few GPIOs. The only options with really serious bandwidth or low latency would probably involve creative(and not necessarily possible, depending on the situation) abuse of camera and screen interfaces.

For all those nice, tractable, problems that behave well on loosely coupled nodes, each individually quite feeble, I guess it'll work; but that certainly doesn't include most of the really obnoxious computational crunching problems.

Re:I suppose...

[ColdWetDog]

Yeah, yeah, yeah. That's all correct.

But you can make a Beowulf cluster out of this.

And that has to count for something.

Re:I suppose...

[lkcl]

Assuming that the obsolete compute modules are of standard size/pinout (or, more likely, that compute chassis are only produced for phones that ship in sufficiently massive volume to assure a supply of board-donors), this scheme would work; but I have to imagine that a phone SoC would make a pretty dreadful compute node: Aside from being a bit feeble, there would be no reason for the interconnect to be anything but abysmal.

the nice thing about a modular system is that just as the modules may be discarded from the phones and re-purposed (in this case the idea is to re-purpose them in compute clusters), so may, when there are better more powerful processors available, the modules being used in the compute clusters *also* discarded... and re-purposed further once again down a continual chain until they break.

now, you may think "phone SoC equals useless for compute purposes" this simply is *not true*. you may for example colocate raspberry pi's (not that i like broadcom, but for GBP 25 who is complaining?) http://raspberrycolocation.com... - cost per month: $EUR 3. that's $EUR 36 per year because the power consumption and space requirements are so incredibly low.

another example: i have created a modular standard, it's called EOMA68. it re-uses legacy PCMCIA casework (which you can still get hold of if you look hard enough). the first CPU Card is a 2gb RAM dual-core 1.2ghz ARM Cortex A7, which as you know is based on the A15 so may even do Virtualisation. i did a simple test: i ran Debian GNU/Linux on it, installed xrdp, libreoffice and firefox. i then ran *five* remote sessions from my laptop, fired up libreoffice and firefox in each, and that dual-core CPU Card didn't even break a sweat.

so if you'd like to buy some compute modules *now* rather than wait for google project ara (which will require highly specialist chipsets based on an entirely new and extremely uncommon standard called MIPI UniPro) the crowdfunding campaign opens very shortly:

https://www.crowdsupply.com/eo...

once that's underway, i will have the funding to finish paying for the next compute module, which is a quad-core CPU Card. after that, we can see about getting some more CPU Cards developed, and so on and so forth for the next 10 years.

to answer your question about "interconnect", you have to think in terms of "bang-per-buck-per-module" in terms of space, power used as well as CPU. a 2.5 watt module like the EOMA68-A20 only takes up 5mm x 86mm x 54mm. i worked out once that you could get something like 5,000 of those into a single full-height 19in cabinet - something mad, anyway. you end up using something like 40kW and you get such a ridiculous amount of processing power in such a small space that actually it's power and backbone interconnect that become the bottlenecks, *not* the Gigabit Ethernet on the actual modules, that becomes the main problem to overcome.

bottom line there's a lot of mileage in this kind of re-useable modular architecture. help support me in getting it off the ground!
https://www.crowdsupply.com/eo...

Wow

[grimmjeeper]

Imagine a Beowulf cluster of these.

Re:Depends on use

[itzly]

Cell phone processors might tend to be slow, but they're rather power efficient per operation

Not particularly. The latest Intel designs are better. Also, the CPU needs big and fast memory, and plenty of I/O bandwidth to do useful stuff. Not typically the stuff you find on a cell phone.

Compute per watt

[itamblyn]

Until we reach a point where compute per watt stabilizes, it is highly unlikely that anyone would be interested in using old components to build a cluster. The fact that the parts would all be slightly different would be a headache too.
Older gear typically uses more power / FLOP, and is slower, so your time-to-solution takes a hit too.
If we get to the point where the power usage / FLOP for an N+1 device is basically the same as N, then you might see people do this, so long as they are okay with waiting longer for a result. Until then, don't hold your breath

Re:Depends on use

[ihtoit]

unless you're talking about a shop in middle Africa or even Outback, China that proposes to utilise such a system in a mesh network to bring remote communities one step closer to being Facebook zombies.

Hell, for that matter - how much processing power do you need to run a DHCP router?
Or a DVR?
Or a home automation system? Something as simple as an automatic garage door opener?
An RFID reader?

There's a BUNCH of uses for low power/small iron that Big Iron would be utterly WASTED on. The aforementioned is not, by any means, exhaustive.

You could also make display walls

[Paul Fernhout]

I suggested this at IBM Research around 1999, and built a proof-of-concept speech-controlled 3X3 display wall of old ThinkPads otherwise destined for "the crusher". Wow, was my supervisor surprised (to put it mildly) when he got back from a two week vacation, as I had built it when he was away so he could not say "no". :-) Another contractor in the lab described his reactions to me though, and helped calm him down. :-)

A couple regular employees associated with the lab had helped me get the equipment. Every laptop had to be officially tracked with an owner and even locked down to comply with IBM policy, even though they had been discarded/scrubbed and were heading for destruction. Ignoring time costs, the laptop locks were the most expensive part of the project in a sense given pretty much everything else was recycled, and a regular employee coworker got them for me out of his own budget (thanks, David!). Another regular employee helped with the networking aspects and tracking (thanks, Mel!).

The people are IBM who dealt with old equipment were very interested in the idea. Who wants to see useable equipment get scrapped? And there was so much older equipment from such a big company, plus from leases and such. But I guess, within Research itself, the project then was not that exciting to people focused on "new" things.

I even wrote up a mock commercial for such display walls with a female executive mother working from home in front of a huge display wall, and her little daughter came by to say hello, and the mom had programmed something fun to show up on the wall for her daughter.

Before we got treadmill workstations, my wife also liked the idea as a way to keep fit -- that you would be walking around all day in front of this display wall you were talking to, rather than sitting in one place and typing.

ThinkPads were interesting in that they could fold flat, so you could layer them on top of each other. However, I also suggested back then that ThinkPads could eventually be designed for reuse in this specific way.

But as just a contractor, and about then hitting the 1.5 year limit for contractors at IBM Research (a rule to prevent them being ruled as employees), the idea sort of fizzled. There was some preliminary negotiations about hiring me as a regular employee, but I probably asked for too much as I had mixed feelings then about the all embracing IP agreements that IBM had and similar things (although I really liked the speech group -- great people), and I also had hopes to even then get back to educational and design software my wife and I had been writing. I did go back a couple more times at IBM as a contractor, but it was for other groups unrelated to speech. Anyway, so that idea faded away.

The display wall looked a bit like part of a Jeopardy set, and you would tell it what specific screens you wanted to do what with. Another speech researcher asked me to set it up in a new lab when I was leaving. So I can wonder if, indirectly, the idea floating around sparked something at IBM Research eventually related to Watson and Jeopardy? :-)

My major use case for the wall was to use as a design tool to make complex engineering projects, like a self-replicating space habitat. However, I also tried to get the IBM Legal department interested in using such a speech-activated display wall for reviewing legal documents and tracking cases, with using such systems backed by a supercomputer becoming a perk for IBM lawyers, but also did not get far with that.

I'm now past the expiration of my non-disclosure agreement on such things that I did or learned at IBM Research back then, thankfully! :-)

Anyway, one could probably do much the same with discarded cell phones...

Epic power wasteage

[GameboyRMH]

The big problem with building a cluster out of anything but bleeding-edge processors is that the flops-per-watt is going to suck so much compared to a new cluster, that you might not save any money over buying that new cluster.

Re:Oh Boy!

[gweihir]

That is not a PC. That is an embedded ARM system. And really, there is no problem with the PC industry. The days of growth are over, but that is _not_ a problem and everybody sane did expect it. A far smaller PC industry 20 years back managed to have several manufacturers for each component and several models for each and prices where comparably lower than today.

Because this is so completely unlike...

[kenh]

...the Raspberry Pi board, you know, that $30-35 "PC" that not needs a keyboard, mouse, SD card, TV, case and power up ply to be usable as a desktop...

What makes his project not cost efficient IMHO is going to be the collection and testing of recycled modules.