backtop


Print 18 comment(s) - last by Moishe.. on Dec 10 at 1:34 PM


Materials discovered using the new cloud computing effort from Harvard and IBM could be used to create flexible organic semiconductor cells able to be printed and woven onto bags or clothing to provide power for gadgets.  (Source: Martin LaMonica/CNET Networks)
Joint project looks to exploit the power of computing clouds to discover new useful solar materials

The use of cloud or distributed computing models is well-established, but still a growing field.  Projects like Folding@Home have taken off, allowing researchers to gain the idle computing power of thousands of volunteers to help crunch the numbers on tricky problems.

In the realm of chemical research, one of the most active fields is solar power.  Much ongoing research is devoted to creating better chemical compounds for solar panel material and even developing organic semiconductors which could make flexible panels with good yields.  However, in order to pick the optimum molecule millions of separate molecules must be individually assessed for their potential capabilities.  This requires a vast amount of computing power, making such efforts prohibitive in the past.  Thus, past solar materials research has largely been a process of slow progress by experimentation.

A new IBM and Harvard University partnership aims to bring solar cell chemistry up to speed technologically.  The new program harnesses the power of grid computing to evaluate numerous organic molecules for their potential as organic semiconductor material in solar panels.  The program is part of IBM's World Community Grid project, an initiative which lends support to many non-profit projects seeking to answer humanity's most pressing problems.  The project already has ongoing cloud computing efforts to find new AIDS drugs and to develop more nutritious rice.

Organic cells are lighter than silicon cells or other inorganic semiconductor cells.  They are also flexible and could be printed cheaply using organic inks containing the semiconducting compound.  Also, some organic materials can absorb a broader spectrum of light than silicon-based cells, allowing them to produce power even indoors.  However, they currently are extremely inefficient and break down far too quickly for commercial use.

Harvard researcher Alan Aspuru-Guzik says the new project will use multiple volunteers' computers to cut down the computing time to analyze the simulated solar efficiency of various organic candidate molecules.  He said that without the cloud's help the project would take 22 years, but with the power of distributed computing, it will finish in just about 2 years.

He states, "It would take us about 100 days of computational time to screen each of the thousands of compounds for electronic properties without the power of World Community Grid."

IBM itself is very much vested in solar research, too.  It has an ongoing program to research thin-film solar cells using CIGS (a combination of copper, indium, gallium, and selenide) and a program to build next-generation solar-concentrator cells.  It also has been developing smart grid power usage software for businesses and utilities.



Comments     Threshold


This article is over a month old, voting and posting comments is disabled

Pay me for my CPU cycles
By Suntan on 12/8/2008 12:39:44 PM , Rating: 3
Not directly related to the article, but I wish my ISP offered a setup such that they would pay me to use my networked computers during idle time. Basically, they could setup a network like F@H and then rent the processing power out to companies looking to crunch a lot of numbers.

Based on the amount of data my computers worked on during the month, they could give me a rebate back against my monthly bill.

-Suntan




RE: Pay me for my CPU cycles
By steelincable on 12/8/2008 12:52:33 PM , Rating: 2
The idea is to use your unused/idle CPU cycles for something useful that would otherwise go to waste. It's a win/win situation. They get free CPU time and you don't 'waste' any CPU time. If they pay you then it isn't really win/win and they would probably just buy more servers than pay you. Could you imagine the administrative nightmare of keeping track of the thousands of users who help them?


RE: Pay me for my CPU cycles
By jRaskell on 12/8/2008 1:09:21 PM , Rating: 5
You're paying your own electric bill, so in effect them using those unused CPU cycles is indeed costing you money, and it could be a lot more than most people realize.

1. I don't leave my home computer powered up 100% of the time. The difference between my current usage (which over-estimating is about 5 hours a day), and a situation where it's running near 100% 24/7 is almost 20 bucks a month.

2. Even if I did leave my home computer on 24/7, the difference between idle power consumption and 100% cpu usage power consumption still is not insignificant.

So, it currently is NOT a win/win situation, since those unused CPU cycles are by no means free at all. Whether the cost is worth it or not is a personal decision.


RE: Pay me for my CPU cycles
By Suntan on 12/8/2008 3:42:31 PM , Rating: 2
Well, I said I wished they “offered” a setup. I didn’t say everyone “should be required to use it.”

First, I have a media PC that is on from 5pm to midnight every night so it is available to record and playback TV shows. Neither activity takes all that much of the CPUs time. So it is there and available. I also have a laptop going a good portion of the time for surfing while watching TV, surfing web pages doesn’t take too much power either so it is mostly available.

As for added power consumption, yeah that’s there, but that would be part of them paying you for the access to your CPU cycles.

-Suntan


RE: Pay me for my CPU cycles
By Moishe on 12/10/2008 1:34:19 PM , Rating: 2
I'm basically in the same position as you. HTPC, multicores and other computers. Some always on. I'd sign up for that.

I actually did a test once to see what kind of power savings I could get. I went from 100% uptime napster or gaming box to just having it on when I used it and I saved $20/month.

Even recovering a portion of the electricity cost would be nice and science would get something out of it. I'd sign up for that.


RE: Pay me for my CPU cycles
By Denithor on 12/8/2008 1:16:46 PM , Rating: 2
Except that it's by no means "free." While crunching all that data your computer typically pegs either its CPU or GPU (or both, depending on the specific client you're running) to 100% and leaves it there. Meaning your system consumes a lot more power than if it was left in an "idle" state - or better yet, standby/hybernated or turned off completely.

Now, don't get me wrong - I'm a Folder myself and actually have a dedicated box with three GPUs just for running F@H. Since the new GPU client launched I've racked up >2 million points on my three folding systems (look up Denithor at F@H).

I do like the first thought there, getting paid for the crunching results would be great. Could help to offset the cost of power & internet access. But instead of the ISP paying I think the company seeking the research should foot the bill. In the case of this particular research, there will obviously be a big financial profit made by the companies involved if/when the research comes to fruition (if they find a way to make solar panels for 50% less that gather 50% more energy -- you figure out what that's worth).

In the cases of disease cure research I think it's more about helping humans in general rather than a financial gain so I'm glad to offer my equipment to the cause. But for a for-profit research project I seriously think they should pay for the resources we provide for their use.


RE: Pay me for my CPU cycles
By Suntan on 12/8/2008 3:49:54 PM , Rating: 2
quote:
But instead of the ISP paying I think the company seeking the research should foot the bill.


Well, the idea of the ISP managing it is that they already have access to all the computers and would be best situated to communicate and deal with the “cloud” as it is. There are a lot of times where companies only want to rent other people’s super computers for a day or a week to complete a specific task, but have no desire to actually maintain one or pay for long term usage. In this instance you would not see them bother with soliciting and setting up their own group of networked PCs, but they would be willing to pay to rent the service from a 3rd party like an ISP.

From the ISP user’s standpoint, they wouldn’t have to deal with other people, it would be streamlined as part of your monthly ISP bill.

-Suntan


RE: Pay me for my CPU cycles
By Bruneauinfo on 12/8/2008 4:14:52 PM , Rating: 3
grid computing exists because we donate our computers. if we didn't they wouldn't build super computers to replace us, they just wouldn't do the research at all. they can do the research our CPU's provide because we make it affordable. F@H works because it is economically feasible for everyone involved. i don't see Stanford buying a petaflop computer when they can use that money to pay for research projects and use our computers for free.

surely someone here can say what i'm trying to say better than i can, but i hope you get my point.


RE: Pay me for my CPU cycles
By Suntan on 12/9/2008 1:26:31 PM , Rating: 2
First, take a step away from F@H. There are other uses for large number crunching than just F@H. While it is understandable to donate computer resources to a good cause, there are a lot of people on this planet that do things because it makes them money, not just because its solely a “good thing to do.” If you don’t believe me, go look up the words “work” “job” “paycheck” “economy” “capitalism”, etc. etc. If that still confuses you, climb up out of the basement and ask your mommy and daddy why they leave the house every morning and they don’t get back until 5pm.

FEA and CFD of new cars takes a lot of power and, while the likes of GM and such can afford to run their own server farm to crunch the numbers (although a lot of them rent computer time from 3rd parties that run their own crunching farm) smaller companies (Tesla, Saleen, etc. etc.) could benefit from more options between the two extremes.

Not to mention when Toro designs a new lawnmower or Trek designs a new bike, FEA is used, but mostly likely is not used to the extent it could be because there is not enough time to do it when you have to wait a half hour for an individual workstation to run through each run.

Having a cheaper option, renting the simulation time from a 3rd party, can reduce the cost and overhead allowing for more simulations to be done in less time.

Same principle as a time share for a Condo on the beach that you only use 5 days out of the year.

In fact, there are currently places that rent out computing time to companies that don’t have the need to justify running their own farm. They just own the computers within there farm. This idea would possibly allow for an even less costly (although admittedly less dependable) option because the ISP does not have to pay for the upkeep of the customer’s computers that it uses.

-Suntan


RE: Pay me for my CPU cycles
By Bruneauinfo on 12/9/2008 3:13:38 PM , Rating: 2
do the math.


RE: Pay me for my CPU cycles
By Suntan on 12/8/2008 3:36:32 PM , Rating: 2
quote:
Could you imagine the administrative nightmare of keeping track of the thousands of users who help them?


Could it really be any bigger than the financial nightmare of keeping track of all the customers that use their internet connection?

Any harder than the phone company billing you per long distance call made? Cell phone company tracking all your minutes as well as your individual texts? Cable company tracking each one of theose PPV movies?

An itemized bill of services is an itemized bill, whether there are credits on it or not.

-Suntan


RE: Pay me for my CPU cycles
By Bruneauinfo on 12/8/2008 4:16:37 PM , Rating: 2
the only reason your broadband service would undertake such a project would be if they could make a substantial profit from it. they are a business and not a charity.


RE: Pay me for my CPU cycles
By Denithor on 12/8/2008 11:33:19 PM , Rating: 2
Same for IBM, which is exactly why I said I would want to get paid for my crunching in a commercial venture like this looks to be. I cannot fathom IBM searching for a more efficient solar panel composition out of the goodness of their corporate hearts...more likely they want the proverbial pot of gold at the end of the solar panel rainbow. And it's a mighty big pot, too, as much interest as there is in solar energy these days.


RE: Pay me for my CPU cycles
By Suntan on 12/9/2008 1:08:14 PM , Rating: 2
From what I wrote, where did you get the impression that I ever thought the ISP would make no profit from it?

The ISP charges Spacely Sprokects X dollars to run its simulation of its newest sprocket. It then farms the workload out to the customers that have signed on to the plan, at the end of the month it credits those customers for the compute cycles they used and keeps the leftover from X as profit.

Pretty much the same thing that certerlized "supercomputer" centers do to generate revenue.

The business side of it isn't that hard to unsderstand here guys. You've never rented anything from someone before? When you only needed it for a limited amount of time? you don't buy a $20,000 bobcat when you only need to move dirt for one weekend, you rent it. Same thing here.

The only difference is that the ISP does not directly "own" the computers it is renting out. Instead it is an intermediary that rents out access to it customer's computers and trades its customers a reduced internet bill in excahange.

-Suntan


RE: Pay me for my CPU cycles
By Bruneauinfo on 12/9/2008 3:11:22 PM , Rating: 2
from what i wrote where did you get the idea that i thought you thought your ISP would not make money from it?

my point is that i don't believe your ISP would make money from it because i don't think IBM would pay them for it because i don't think IBM's customers would have the money to pay for that kind of service.

it is because folding@home is 'free' for us to download and use that it makes economic sense for research. otherwise they would spend their research dollars elsewhere. the cost of running a supercomputer of equal power would be very cost prohibitive. for example: the folding@home grid supercomputer constantly updates itself with newer and additional hardware, when a node goes down the IT department doesn't have to fix it, no power bill to operate the computer or to cool the server farm, no buildings necessary to house the computer, no taxes on the computer... i'm sure there's more....etc

that the computer is constantly upgrading itself is reason enough that they could never afford to purchase such a machine. a petaflop computer today already costs a great deal to build and operate, much less update with new hardware every few weeks. a grid computer will constantly upgrade itself which is very costly to replicate.

you want to say "well, that 'very costly to replicate' warrants my ISP and IBM getting in bed." however it doesn't, because IBM's customers can't afford to pay the alternative cost in the first place. the cost of keeping a super computer is cost-prohibitive. a free grid computing system on the other hand makes economic sense - makes things economically possible.

for example: say a 50,000 unit system grid at 1 cent per day paid to each participant. do the math. it would not be cheap. folding@home is larger.


"This is about the Internet.  Everything on the Internet is encrypted. This is not a BlackBerry-only issue. If they can't deal with the Internet, they should shut it off." -- RIM co-CEO Michael Lazaridis














botimage
Copyright 2014 DailyTech LLC. - RSS Feed | Advertise | About Us | Ethics | FAQ | Terms, Conditions & Privacy Information | Kristopher Kubicki