This is a public network of computers. That will make money.
Just about every company needs IT infrastructure for there projects. Home users need it but usually can not get access to it. Because of the high costs when dealing with company grid computers. This system uses public grid computers, so costs are dramatically derived down.

I am new to this site, but it sure seems interesting.

LOVE IT

I like this idea and would certainly like to delegate some of my video encoding tasks and such like to it. Does anyone have the technicalexpertise to get it up and running though? I know I don't.

My idea CrowdProcessor was somehow alike. I think we need a good developing team who can convert usual applications to parallel processings. I would love it however!

this has some convergence with this (http://www.cambrianh...ss/view/renderjuice/ )

Nice idea but I have no idea if it is possible (doubling words today...)

AWESOME

"Does anyone have the technical expertise to get it up and running though? I know I don't."
Personally I have set up small in-house networks that work in groups. It is possible to have this project up and running with the right group of people.

"this has some convergence with this"
The ideas are similar as far as having a computer farm. That part of the idea was never new, computer farms all ready exist. The main point is that this would aim at other applications, target new user groups, and have an entirely different business model then all other services.

Q: Whats the business model?
A: Selling the extra computer power to contributing users that want extra bandwidth or to non-contributing users that would like to buy the power.

There are lots of companies that have projects that require massive amounts of processing power. On there own a project would take them months before they could ever get it produced. In addition, there costs would be through the roof.

Thousands of companies would gladly want to participate in this program or buy the extra computer power. Because we could sell it at a very low competitive cost. Since, all of our hardware contributed and shared by users of free.

Users that contribute and join the system would get to use it for free because they contributed more processing power. As more and more users contribute the more the system would be improved and we would make even more money from that extra possessing power.

How contributing users can have this service for free:

1.) The users signs up and downloads our networking program to the computer.

2.) They select when they want there computer to connect to our network. Basically when every there computer is idle or not doing much it connects to the network.
For Example:
* Instead of a screen saver our program would connect the user to our network and the computers idle time would be used.
* If the computer was left on and idle our program would connect the user to our network and the computers idle time would be used.
* Simply set the amount of power he wants to contribute in the background while still using the computer.

3.) For there participation they get to use the system free for there personal or commercial use. How much they can use the system is allocated by how much they contributed to the system.
For Example:
* How fast there computer can work.
* The speed it can connect.
* How much time they connected to the network.

So basically contributing users get to use super computer for what ever there hearts desire at no cost!

i like it!

This idea is amazing, and would rake in a fortune.

This project could benefit from my 'Project Aldeburgh', coming soon to CH. You can read about Project Aldeburgh at http://www.craighenderson.co.uk

I'll post an Idea here for it soon.

I assume it could benefit a lot of projects. Granted they will need to be to the right scale to stay efficient.

Quote From User Drh:
------------------------------------
As well as allowing members to get free use of the super computer they could have the option of taking their reward as cash or time. E.g. at the end of the month they have earned 200 credits which would entitle them to and hour of computing power or $10 which ever they prefer. The cash payout would have to be based on how much of the processing power had been sold that month, kind of like an exchange rate.

I was thinking that students would be a great member market to target as I remember when I was at Uni everyone had PC's that were left on all the time and we always need a bit of extra cash. University departments also might be good customers of the computing power, science departments, computer departments etc.
------------------------------------

This idea is not new at all.
It's the same a my p2p render farm idea from the last voting rond

]V[oogy

1. This is NOT p2p.
2. This idea was originally posted a long time ago by myself. (I re-posted it for clarification and to address every ones concerns and opinions.)
3. This idea uses a different business model.

seti at home but with profit?

Yes, it is kind of like seti at home. The difference is this is not just for non-profit organization projects, it is for regular companies and even for personal use.

The contributing users users of the service can donate, use the farm for there personal or commercial use, or make profit. Unlike seti at home which is only for major non-profit projects which offers nothing to contributing users.

My idea is not just lets build a super computer, lots of people have done that before. My idea is to use the computer farm to produce both noncommercial and commercial endeavors.

This idea sounds very good but it is very terrible for the users. Using 'idle cycles' doesn't sound very bad, but it isn't good for the computer.

The users will most likely be spending more in power than the money they would get out of it, and their computers would be shortening their life time greatly.

Check out my Idea, specialfx@home its very similar in some areas and the suffers from the same problems.

I like this it has allot of technical detail that would need to be addressed.

Just like with my idea, I believe the necessary skill set may not be here.

Many users would have been leaving the computer on anyway. With the monitor turned off, the average computer uses less electricity than a light bulb, even less for laptops. Even if you left the computer on all day it would cost you at most a few pennies a day.

You will not see any harmful happen to the hardware that is extremely unlikely.

Also, keep in mind that users would be gaining from contributing to the service. Because they can use it themselves, revive money or make a donation to help others.

This seems like an idea which is worth exploring to see if it is feasible.

If I were an investor I would say, "Come back with a proof of concept".

Since we are all talking about ideas here, I think it's fair game to have untested / unproven ideas. A positive vote indicates you should take it to the next step.

Intuitively I would shy away from signing up for a system like this. I am a non-techie. Anything about "sharing" my computer suggests vulnerability to viruses, hackers, identity theft & things that go bump in the night. I bought my own computer so that it would be my own, period.

However, it might work if you could created trusted neighbourhoods or syndicates of people in closed groups who only share with each other. Maybe.

Personally I don't think that this system could pay me enough to put my perceived privacy at risk. I would be highly unlikely to sign up to use something like this. A large part of that is because I don't understand how computers work and I fear what I don't know. I think many people are like me in this regard.

It would not matter how many experts say that my system is not at risk. I have no incentive to share my system so I would not sign up for a sharing service like this one.

Who do you think WOULD use it? What demographic?

so you stole Moogy's idea?
I think this has been done before. P2P is nice and dandy, but when it comes to those people actually paying, good luck! Another issue you will find that upstream rates are pretty poor, for eg in Bell's megabit service, the upstream is only 16 kb/sec.

I did not take Moogys idea, in fact I posted this idea way back in 2006 months before Moogy had posted his. My idea was also posted long before P2P. I re-posted my own idea, to clear up my thoughts and try to communicate clearly to the community. Feel free to look on my profile of ideas.

"Who do you think WOULD use it? What demographic?"
The main demographic groups are:
Movie Makers
Computer Artists
Computer Science
Mathematicians
Medical Researchers
Scientific Researchers
Most Companys

There are also lots of other groups like:
Computer Gamers
Website Owners
Just about anyone...

There are not very many limits to who can use the network.

Make it happen! Let's make this happen. I like the idea of any idle resources (of the affluent) being shared to those who lack these resources. I read about something like this during the debate about the Sun Microsystem paradigm. I dont know what happened. But definitely it did not reach the PCs of the ordinary people.

I give you four on this Kevin!

hmmm...very good and would really help a lot, including my own projects. I dont care who originate this as long as ti can be done thing and much better than the "first" still is;
sweet and Rawkin!

GET GOING! it make lot of sense...

This idea with respect to purely 3D graphic rendering farms (public ones) has been done and is underway with the BURP project in alpha :

http://burp.boinc.dk/

All of it is free, and publicly available, from the free open source 3d content creation suite (yay BLENDER) to the renderfarm itself ...

More info:

http://www.blenderna...roject-enters-alpha/

http://www.blender.org/

Of course, you could promote your idea to all the folks out there not using Blender, as BURP seems to be specfic to Blender only...

They are not making money from it they are only non-profit projects.

I use Blender, whats wrong with promoting Blender with this project? 3D artists need money to fund there projects. It seems like a good trade off extra computer idle time to money.

Using a network of computers around the globe is a great idea.

I am sure, however, that everyone who would even contemplate joining, would be frightened silly by the possibility of HACKERS joining the scheme and frying a few hundred machines just for laughs.
I am certain that you have already explored this possible hole and plugged it.

I can think of some negatives, but I am willing to shove them into my back pocket and forget about them.

Its just too easy to be negative - so I am giving it a BIG thumbs up!!!!

Well, I can't say that there is never going to be some security hole. If someone really wanted to they probably could find a way. But, the same is true for all systems and every computer on the internet.

Over all because this is more of a controlled system, it would be typically safer then the browser you're running right now!

Thanks for answering my questions, Kevin. For the type of users you are aiming at, and based upon the feeback here, it seems to be an idea worth pursuing. Good luck!

First, I think this is a great idea. However, one big concern I'd have if I were an investor would be "How can this compete with a service like Amazon's Elastic Cloud Compute service?" ECC is basically the same principle as this idea, except instead harnessing unused cycles from users, it takes the excess computing power from their huge data centers and sells it to users who need processing power. The rates are extremely low (.10 $US/hr) and they guarentee high reliability, security, etc..

A non-profit distributed p2p service would be great, but a for-profit project would need to have razor thin margins to compete with Amazon in this niche.

$.10 is there flat rate for just the CPU, they have other charges that apply are also applying.

On average a server farm charges with all costs around $.40 to $1 per hour. Also, you have to keep in mind they have to pay for the storage, maintenance, the hardware, everything. Also, server computers are have slower processors then the most home computers.

"guarentee high reliability, security, etc.."
This service would be very redundant system, that would maintain reliability. Also, no one can "guarantee" security on any computer system. Not home computers, not major corporate websites, not the government, no one.

Wont this be another pipe dream? Looks like a very HUGE job as in HUGE!...i knew wayback this has been in the works, but even moneyed corporation failed to lift this off half inch above the ground.

It s the collaborated open source 'rendering" that seems the big block? I dunno if am right am more into vitamins and viral infections=)

Honestly, its cool!

The technology is available. Non-profit server farms like this have been set-up successfully. There has been open source non-profit render farms as well. This idea basically changes the business model to turn a profit as well as bring it closer to the point where the average user would have access.

Honestly this is an idea and all ideas are risky. You have to take a change to see if it will work. All, inventors and entrepreneur have to take risks before they can ever turn a profit or see if something will work.

FocusDriven, actually, I don't think it'd be a big deal to write. Think of a program similar to a torrent client - it would go out to a supernode and find machines to talk to. It would then negotiate a work unit to be processed with each willing (registered for payment) node. The trick is for security to be paramount, the work unit would likely need to comprise some sort of executable (native, java or other bytecode machine) that could be run in a sandbox as not to jeopardize the host (unless of course the host ran a signed fixed function binary). Data would need to be encrypted for transport, and would be susceptible to examination, as the node needs it in the clear to operate on it. A sandbox, VM or similar approach would provide the safety and flexibility required. Some issues crop up - not all processors have the same SSE extended instructions, so it means some things will not be as fast - and the whole point is to maximize FP right? As well, any Java or interpreted program will not run as fast as a binary, and you'd need several flavours of binaries to encompass the various CPU types (like Seti@home does). The technical part is NOT the problem. The problem is how to publicize, get people to participate, and to offer an incentive for people to do so ( money!). The node client will need to be closed, so that it is not hacked, and I expect people will be very wary that their machines could easily become hijacked or operate as spambots. Good luck with 10c paypal and VISA transactions with 75c-$1.50 overheads!

I assume you got $.10 from amazon. But, you are slightly confused because they have other charges that apply. Most farms charge about $.40 to $1 per CPU hour.

Server processors are generally slower and they are not set up to process graphics. The reason being is because servers don't need to process GUI. Home computers have much faster processors and GUI processing capability.

The code data being sent would not have to necessary run any executable code. This system would be safer then the browser you are currently running right now.

Also, the user has options they can use the time, receive payment or donate it non-profit.

The payment system would probably work much like most ad services, that is the payment would be sent every $20-$100 dollars. Over all the users is not going to notice any downfall because this is idle time anyways.

bad ass idea

Kevin and all, I'd like to raise a layman's question:

The 24/7 300 kbps internet cost here in the Philippines is around US$20.8 a month. If I have a PC with speccs: Pentium 4 with 256 MB RAM, 1.5 ghz. What is the possibility of recovering the internet cost with this? Up to what percent?

My fervent wish for projects like this is it can be a cost recovery scheme for people incurring expenses on internet connectivity especially for those living in a tight budget.

This idea can rock the world Kevin.

Love the idea.

Maybe apple will buy it and add it to their .mac subscription. Possibly for pros only, (Final Cut Studio Users), just a possibility.

"The 24/7 300 kbps internet cost here in the Philippines is around US$20.8 a month. If I have a PC with speccs: Pentium 4 with 256 MB RAM, 1.5 ghz. What is the possibility of recovering the internet cost with this? Up to what percent?"
Well that would depend on how much time the computer is left idle.

Bare in mind this are rough estimates power consumption. The math is correct, but the power used by each users machine will depend on the hardware setup of the machine.

Lets say you have 6hours of computer idle time a day, this would be easily achievable by anyone.
Assuming you turned off the monitor, with the computer using the maximum a full load. (Note: A full load is the MAXIMUM cost that should be seen to any user. A average computer will operate at about 80% efficiency due to the nature of the circuits.)

For 6 Hours you computer will use about 2.256 Kw
Assuming it costs 10 cents per unit of electricity the cost of running for 6 hours is about 2.256 * 10 = 22 cents or less per day.
22*30=660 cents or $6.60 per month.

Now to calculate the profit you will make. c=cents
If you make 5c per cpu hour for 6 hours:
(5c per cpu * 6hr) - 22c = 8c per day
8c * 30 = $2.40 per month

If you make 10c per cpu hour for 6 hours:
(10c per cpu * 6hr) - 22c = 38c per day
38c * 30 = $11.40 per month

If you make 15c per cpu hour for 6 hours:
(15c per cpu * 6hr) - 22c = 68c per day
68c * 30 = $20.40 per month

If you make 20c per cpu hour for 6 hours:
(20c per cpu * 6hr) - 22c = 98c per day
98c * 30 = $29.40 per month

If you make 25c per cpu hour for 6 hours:
(25c per cpu * 6hr) - 22c = 128c per day
128c * 30 = $38.40 per month

If you make 30c per cpu hour for 6 hours:
(30c per cpu * 6hr) - 22c = 158c per day
158c * 30 = $47.40 per month

Also, on another note this is not just limited to computers. Game stations with an Internet connection would work as well. For example you could even use your PS3, Xbox360, Wii.

Kevin,

Reasonable idea.

One challenge that a number of folks have alluded to is the security. Most comments were related to those people donating the processor time. What about the companies/firms you are suggesting could buy this service and truly make it profitable? What is the pitch for a company and how do you sell the business on security, particularly in this day and age of privacy and so forth?

As “theGuruâ€� states, technically achievable; however, the concept needs to bring return. I might well have missed this but I’m wondering â€" who monitors the available processing time and payment to those contributors? Do contributors get paid whether “somethingâ€� is using their available time? In other words, if 100 of us agree to provide 10 hours a day of our personal PC cycles, there is a 1,000 hours per day available. Are all of us getting paid? And if not, what is the mechanism for monitoring/control?

I may be jumping into the weeds a little too deeply here, as I’m a new player to this game. So, please don’t take this as criticism; however, this aspect seems critical to the success of the business concept…

Buytore

Q: Security?
A: Company's can be rest assured that there data will be safe and secure. Even if there was a problem in security the data would be broken up and would be useless to an attacker. Contributing uses also should not fear this system will have security protocols in place. Also, the code being sent would not even have to be executable. Overall like I said before it is much safer then the browser you are currently using.

Q: What about the companies/firms you are suggesting could buy this service?
A: There is a wide range of companies and individuals that need computer data processed for them. Both civil and military departments and agencies. Both companies that do not have the ability to invest in IT infrastructure and companies that have existing infrastructure that need to expand there capabilities.
Some examples: Computer arts, automotive, aerospace, researchers of all kinds, film makers, web designers, financial services, game developers, media analyzers, weather forecasting, climate research, simulations/scenarios, medical, chemical, life sciences, etc...

Q: Who monitors the available processing time and payment to those contributors?
The home server monitors the available time, and calculates payment for contributors.

Q: Do contributors get paid whether “something� is using their available time?
There are many way this can be set up.
Contributors will get credited by the amount of data processed, type of data data processed and how fast the data was processed. The amount they get credited depends on the amount we make for the project.

Many projects will have work that is exponential, but lets say there just happened to be some free time. That time would be donated to non-profit organizations or if the user chooses no data would be processed. Contributors would not receive earnings credited. This is similar to google's psa adverts.

Also, if the user wishes the program on the contributors computer could shut down the system after a certain number of queries that show no work available. Keep in mind that the program under the users control running the shutdown command, not the online server. So, this would be safe from attacks.

Other questions I have received and answers to them -
Q: What about Reliability?
Because the work is not being done by a single group of computers, in a single location. You don't have to worry about the power going out, natural disasters because the system is fail-safe because of redundancy. It is critical to most customers that servers be up and running, no matter what. This service would provide that the best hands down.

Q:What about slow CPS?
The number of CPUs on a render farm impacts performance more than the combined clock speed of the CPUs. Example: Six or nine 1-GHz machines chained together will render much faster than two or three 3-GHz systems.

I like your IDEA, can you give me your email address for some offline inquiry re 3D development of my group?

Just send me a message ok?

Goodluck and congratulations for winning the 3rd weekly tournament!

Well, I have not won the weekly tournament just yet. Only 3 days, 2 hours left to vote in this round! It is not over until it is over.

But, thanks for the message of anticipation for the foresee able future. I will be sure to contact you.

I believe this idea is going to win Kevin! And I think even your closest competitiors in this tournament would be happy to see this idea win. This is a great! It is not just a flash of a genius, the 99% perspiration is also obviously present. Good luck and congratulations!

"Server processors are generally slower and they are not set up to process graphics. The reason being is because servers don't need to process GUI. Home computers have much faster processors and GUI processing capability."

Server processors are NOT generally slow then home machines. Have you looked at any of the current blade based systems?? Opterons, 64 bits, multiple cache levels, scsi disks, gigabit ether. Slower? Actually, a desktop having to run a gui makes it MUCH slower, even if there is an accelerated graphics card, because (in the case of Windows) you have this enormous cruft of drivers and layers of software. Now if you are using an application that specifically uses the GPU processors to do compute, that is another story. Even then, there are several GPUs to support, and only certain models are worth it.

If this is the best idea on CH, that says too much about CH!
The creme is not the only part that floats to the top!

From The Guru: The problem is how to publicize, get people to participate, and to offer an incentive for people to do so ( money!).

From Kevin's most conservative assumptions:

Now to calculate the profit you will make. c=cents
If you make 5c per cpu hour for 6 hours:
(5c per cpu * 6hr) - 22c = 8c per day
8c * 30 = $2.40 per month

My comment: I think the overwhelming crowd support to this idea in this tournament can be extended to the implementation stage.

The 6 hours and 5c per cpu hour assumptions is also conservative in a typical home computer. Further, in the Philippines, the average occupancy rate of Internet cafes is around 5 hours per station in a 15 hour operations. Meaning if they go 24 hours, they can spare up to 19 hours per PC. In a 10 station I cafe, and at 5 cents per hour, the extra monthly income could reach $285 or P12,825. More than enough to pay for the rentals.

In the household, the $2.40 extra income or savings is already a good incentive for people who are in a tight budget. Much more if the income can significantly if not fully subsidize the $20 monthly internet cost.

This idea has great potentials in bridging the digital divide!

"Server processors are NOT generally slow then home machines."
Modern day server clusters use processors that are comparable or often times generally slower then most general home computers.

"Actually, a desktop having to run a gui makes it MUCH slower"
This is true, but you can run things by command line. Also, my point was that server processors do not use the same architecture.

Don't get me wrong servers are great at what they do. The reasons why they are set up that way is because it is efficient. One thing they will have over this service is a faster network connections, because they are set up locally. Which may or may not be that big of a deal depending on the work type.

Another thing to thing about, cooling and storage costs for modern server clusters costs companies a lot of money. Something that we would not really have to worry about.

"The problem is how to publicize, get people to participate, and to offer an incentive for people to do so ( money!)."
Publicize: that part is not hard run a few ad campaigns and a few press releases. Most work will probably be done by word of mouth.

Getting people to participate may be hard at first. Because individual program support for several types of computer setups will need to be made. Also, the people buying the time are probably going to need some standard way of working with there different types of projects. It takes time to get everything up and running.

Incentive: Get to use a super computer/Get paid money.

One of the problems that I think will be the hardest part is getting enough investors to get the service up and running. Because this would still take some investment to set the service up.

Kevin, what you are saying about modern home computers versus modern servers is blatantly wrong. Please do some research. In terms of architecture Opterons and Xeons (The server CPUs from AMD and Intel) are based on the same architecture as the desktop parts but typically have allot more cache, this means they are typically better at all the tasks you propose.
The system architecture is built to have a higher level of redundancy. More robust power supply's, error checking memory and multiple hard disks. Not only are they fast they are dependable.
Also real time photo realistic graphic in games does not make sense. This network would have far too much latency to be practical for games.
Amazon pricing for all interested(main competition imho):
Pricing

Pay only for what you use. There is no minimum fee.

Instances
$0.10 per instance-hour consumed (or part of an hour consumed)

Data Transfer
$0.10 per GB - all data transfer in

$0.18 per GB - first 10 TB / month data transfer out
$0.16 per GB - next 40 TB / month data transfer out
$0.13 per GB - data transfer out / month over 50 TB

Data transfer "in" and "out" refers to transfer into and out of Amazon EC2.
Data transferred within the Amazon EC2 environment, or between Amazon EC2 and Amazon S3, is free of charge (i.e., $0.00 per GB). All Amazon S3 storage and request charges will still apply.

Amazon S3 usage is billed separately from Amazon EC2; charges for each service will be billed at the end of the month.

You have complete control of your instances. You have root access to each one, and you can interact with them as you would any machine. Each instance predictably provides the equivalent of a system with a 1.7Ghz x86 processor, 1.75GB of RAM, 160GB of local disk, and 250Mb/s of network bandwidth.

How do you compete with this?

BTW here is a the valve hard ware survey to give you an Idea of the Typical Gamer(high specified home computer)
http://www.steampowe...m/status/survey.html
Also note, I have used Amazon EC2 and it is fantastic, my Instance CPU spec was much better than the advertised 1.7ghz, I typically had a dual core 2.0ghz spec system!

Servers clusters have a lot of processors the processors in them, but the processors them selves run at about 1GHz-2GHz. High CPU speeds are not as critical for many servers than they are for desktops. If more processing power is required in a server the typically more CPUs will be added rather than increase the speed of the CPU. Why, because in a server you want reliability and redundancy.

Contrary to popular belief, it doesn't take a very powerful processor and ram to simply serve data over a network. It is your desktop that does all the processing for the most part.

Also, servers lack dedicated graphics processing cores.

"How do you compete with this?"
Simple more computers dedicated to the same task. You need to be thinking in teraflops here. The storage can be held on a single location or sent only at specific times. Most of the processing for a lot of projects would not even have to be done online. Rendering data is basically a set of raw commands.

Amazon is basically doing what this service would do, they are using idle machines to process other peoples data. The difference is they have to own all the hardware and pay for a lot more expensive.

If you have ever used typical render farms, you will quickly find that they charge much more then Amazons beta service. The average cost is $0.40 to $1.00 per hour.

"Also real time photo realistic graphic in games does not make sense. This network would have far too much latency to be practical for games."
It could be done, not saying it would be easy. Basically you need a large buffer and a way to send data the data post rendered. This was more of a reach goal.

The reliable of this project is extremely redundant, because there are so many computers in the network, in spread out locations. If one computer goes out the data is simply process by a different computer on the network.

This system does not keep all the eggs in one basket. Even if nuclear bombs, earthquakes, floods, hurricanes, tornadoes, monsoons, meteor strikes, volcanic eruptions, street riots, alien attacks, power outages, you name it all happened at once. I am willing to bet that this network would still be able to stay online and humming. Well maybe not alien attacks if they had some kind of bad-ass dooms day weapon.

As soon as you start paying for licensing of renderer's (which actually does have a real cost), you will have to rise the price of the service, as well as the cost of having a custom client made for you network. Reliability is not really an issue with amazons service. Amazons servers are typically on par with your Average gaming system which is above par compared to your average home PC.
Those specifications are by no means shabby. And render farm CPUS are highspecd. And graphics cards are hard to use for anything beyond realtime graphics. Research Raster vs Ray Traced to learn more about why.

"As soon as you start paying for licensing of renderer's (which actually does have a real cost), you will have to rise the price of the service"
A. Many rendering applications for farms are free. or Can be exported to 3rd party rendering applications. Most existing render farms all do this all ready.
B. If there is a licence fee on a specific product and no way around then that means everyone would have to pay for it. There for included in all costs even for the competitors.

"Reliability is not really an issue with amazons service."
They have reliability, however our level reliability is more dependable. Also, keep in mind they have a very limited load there servers can handle.

"Render farm CPUS are highspecd"
The number of CPUs on a render farm impacts performance more than the combined clock speed of the CPUs. Example: Six or nine 1-GHz machines chained together will render much faster than two or three 3-GHz systems.

"graphics cards are hard to use for anything beyond realtime graphics"
Graphics cards are used all the time for pre-rendered graphics and physics. It is more efficient to run a programs written for CPUs on CPUs and it is more efficient to run a program for GPUs on GPUs.

GPUs on the average performance systems like game, media and workstations have more power then even the CPU. GPUs have increased in programmability and performance over the years.

Also, GPUs can also be extremely useful for non-graphical problems (GPGPU: General-Purpose computation on GPUs).

https://renderman.pixar.com/store/
$995 per machine.

Who says they have a limited load they can handle? This is unsubstantiated.

"Example: Six or nine 1-GHz machines chained together will render much faster than two or three 3-GHz systems."
Where is this example from.
Faster cpus typically have further refinements as they are more modern.
For example Sisoft sandra, a benchmarking suite reports that a modern Core2 Duo t7400@2.16ghz gets a floating point score of 63352, where as an older Amd Athlon xp2800 @ 2.08ghz gets a floating point score of 19391, if we double this as it only has a single core, we get 38782 points. This is 1.63x faster,the clock speed differential of 80mhz hardly accounts for this.
The trend follows with the other benchmarks. This totally defeats your logic. In fact Sisoft Sandra suggests it would take 4.3 pentium 3 1 ghz machines to match the speed of a single AMD Athlon 2ghz machine (The server version is faster again!).

As for general purpose computation on GPUs is definitely not mainstream by any means, only a fraction of home computers could dream of running this setup(folding at home only supports 1 model of GPU(typically one of the most expensive!) for this, which makes up only 6% of the total computing power).

The only commercial renderer I know of that runs on the GPU is Nvidia Gelato at $1500 for the software for shader and network support: http://www.nvidia.co...age/gelato_faq.html.

Sorry to be blunt, but please do some more research.
My own specialfx@home suffers from the same problems.

Everyone wants a teleporter, best idea in the world next to a money tree. But can you make one and where would you start, and how much would it cost top build?
This idea sounds super appealing but the scope is immense to provide a competitive distributed super computing network. Your enthusiasm is fantastic. I recommend you focus the scope of this project more. Choose a particular commercially valuable task that works with the limitations of the product.
Otherwise you have a teleporter.

The BOINC platform http://boinc.berkeley.edu/ already exists and can be utilised for the type of applications you suggest.

Good one.
Check out Boinc and see if your ideals align with theres. What service can you offer on top of this?

From the BOINC pages:
Low data/compute ratio
Input and output data are sent through commercial Internet connections, which may be expensive and/or slow. As a rule of thumb, if your application produces or consumes more than a gigabyte of data per day of CPU time, then it may be cheaper to use in-house cluster computing rather than volunteer computing.
What resources are needed to create a BOINC project? ¶
If you have an existing application, figure on about three man-months to create the project: one month of an experience sys admin, one month of a programmer, and one month of a web developer (these are very rough estimates).
Once the project is running, budget a 50% FTE (mostly system admin) to maintain it. In terms of hardware, you'll need a mid-range server computer (e.g. Dell Poweredge) plenty of memory and disk. Budget about $5,000 for this. You'll also need a fast connection to the commercial Internet (T1 or faster).

"Who says they have a limited load they can handle?"
There is a physically limited number of servers they have.

"Core2 Duo"
There is a difference between actually having 2 processors and having Duo Core.

"only a fraction of home computers could dream of running this setup"
Yes, there would be limited support for only some models of GPU with fast GPU performance systems like game, media and workstations.

"Otherwise you have a teleporter." No one has built a teleported before, multiple groups have built networks like this before. This idea is not to reinvent the wheel. The ideas is to become the next Goodyear or Michelin. The people making money from the wheel.

Did I say this was not going to cost money to start up or maintain?
NO, I did not and quite clearly stated that it would still take investment. Just not as much as typical setup operations.

Did I say there were not other projects that made supter computers before?
NO, I did not and quite clearly stated that that part is not new and that there were several other projects that made supercomputers over a network.

Did I say the network connection was going to be faster then a directly networked local lan?
NO, I did not and quite clearly stated that that was going to be a slight disadvantage depending on the type of data being processed.

"There is a physically limited number of servers they have."
Just as there will be a limit to people on your network, these numbers are both free to grow.

"There is a difference between actually having 2 processors and having Duo Core."
Yes, a dual core processor is typically faster. Read the benchmarks!

"only a fraction of home computers could dream of running this setup"
Yes, there would be limited support for only some models of GPU with fast GPU performance systems like game, media and workstations.

"Did I say this was not going to cost money to start up or maintain?
NO, I did not and quite clearly stated that it would still take investment. Just not as much as typical setup operations." I would say more so. My ec2 setup is costing a fraction as much.

"Did I say the network connection was going to be faster then a directly networked local lan?
NO, I did not and quite clearly stated that that was going to be a slight disadvantage depending on the type of data being processed."
In fact this disadvantage is so great that it makes movie rendering impractical for anything but hobbyist applications, Also as the Guru mentioned, you would soon be violating your home internet contract as well if you used it commericially which is precisely what you are doing if it isn't for hobbyists(ie at no charge!).

"Just as there will be a limit to people on your network, these numbers are both free to grow."
There is quite a big difference between numbers. Also, they own all of the hardware it cost them money to expand. The reason why they have ec2 is because they don't want there systems sitting idle costing them money.

"Yes, a dual core processor is typically faster. Read the benchmarks!"
No, you are misunderstanding something. You are talking about a different structure. Duo core has to CPU on the same chip that share the same slot. Dual Processor is having two separate CPU chips each with its own slot.

Also, Core2 Duo means it has 2 duo core CPUs AKA 4 CPU not 2! Next, a duo core 1 Ghz isn't equivalent to a 2 Ghz either. Even in Intel's own literature says there is only 40%.

Your facts are way off.

"violating your home internet contract" No, you are misunderstanding the contract. You can't run a server commercially from your home connection. The server is on our side so there would be no violation. Next, for most projects the processing is done almost entirely off line.

"Also, Core2 Duo means it has 2 duo core CPUs AKA 4 CPU not 2!"
I am sorry but this is completely wrong. There are only 2 cores in a core2Duo, period!
I have given you links to facts, read them!

OMG...
Core Duo (2 Processors) + Core Duo (2 Processors) = Core2 Duo (4processors)

What maybe I am thing about Core 2 Quad.

I do not wish to spoon feed you but read this.
http://www.intel.com...od_desk+sc_processor

Ya, I was thinking of Core 2 Quad which has 4 Processors.

Anyways back to the point, the number of processor chips matters more then the clock speed of the chip.

That is not what you said. You are changing your argument.
You said:
"The number of CPUs on a render farm impacts performance more than the combined clock speed of the CPUs. Example: Six or nine 1-GHz machines chained together will render much faster than two or three 3-GHz systems"

And as per the numbers I showed you,

Core2 Duo t7400@2.16ghz gets a floating point score of 63352, where as an older Amd Athlon xp2800 @ 2.08ghz gets a floating point score of 19391, if we double this as it only has a single core, we get 38782 points.
Data from SiSoftSanda http://www.sisoftware.co.uk/
1.63x faster!

That is different how?
"The number of CPUs on a render farm impacts performance more than the combined clock speed of the CPUs."
"The number of processor chips matters more then the clock speed of the chip."

A. Your not even comparing processors of the same brand. They don't even have similar specs and there is a vast age difference.
B. You are talking about 2 Processors on the same chip, I am talking about processors each with there own chip.
C. You can't just add processor speed up that way.
D. What exactly are you trying to argue anyways?

Read your own quote:
"Example: Six or nine 1-GHz machines chained together will render much faster than two or three 3-GHz systems."

A. Your not even comparing processors of the same brand. They don't even have similar specs and there is a vast age difference.
Thats what I am saying, newer chips are faster clock speed(ghz) for clock speed(ghz).

B. You are talking about 2 Processors on the same chip, I am talking about processors each with there own chip.
Chips with multiple cores tend to be newer and faster. There is no notable advantage to having one cpu per chip except for thermodynamics(less heat density makes it easier to cool those cpus).

C. You can't just add processor speed up that way.
You cannot add the GHZ to get a meaniningful benchmark, hence why i sent you computational figures from Sisoft. These calculate actually CPU work done(which has real meaning) regardless of CPU cores or clock speed(ghz)

D. What exactly are you trying to argue anyways?
First, being that 3 modern CPUs @ 3ghz beat the crap out of 9 old @ 1ghz ones. Because new chips do more WORK per cycle!
Secondly,I guess it comes down to the fact that you are inconsistent with your facts and do not have the relevant knowledge to even estimate the scope of a project like this or its practicality.
I think its a shame that this idea(as it stands) is a pipe dream and more executable and practical ideas will miss out on the opportunity to win this weeks idea wars.

Best idea I have seen on this site. Keep up the good work.

The number of CPUs on a render farm impacts performance more (Six or nine 1-GHz machines chained together) than the combined clock speed of the CPUs (two or three 3-GHz systems).
They both mean the same thing.

Thats what I am saying, newer chips are faster clock speed(ghz) for clock speed(ghz).
So, what your saying you proved this great mystery. That old chips don't have the same clock rate as new ones.

"These calculate actually CPU work done(which has real meaning) regardless of CPU cores or clock speed(ghz)"
Yes, but it only calculates single processors work done with the processor isolated and not sharing work. This is not calculating clusters of farms.

"Because new chips do more WORK per cycle!" But, they don't split the work up as much. It is all about how the work gets shared. That is what makes it faster. If you have ever set up a render farm before, it is easier to grasp.

Like many ideas ... this one sounds good until its time to implement it. I do believe its possible, i even believe some people would be happy to pay for such a service.

the questions involve HOW:

How will it work?
How will it be distributed?
How will you charge/collect money?
How will you distribute rewards (time/money)?
How will you market it?
How will customers take advantage of processor time?
How will you manage multiple project requests for processor time?

The more of these you answer, the clearer the opportunity (or not) will appear.

"How will it work?"
Basically like other super computers and render farm projects. They down load are client and uses there systems idle time to process other data.

"How will it be distributed?"
The inter-web.

"How will you charge/collect money?"
The same way other online services charge and collect money.
"How will you distribute rewards (time/money)?"
The same way other online distribute awards. With this system you will actually earn credits that can be redeemed for money/time. Similar to C$ on this system.

"How will you market it?"
Ad campaigns and a few press releases. Most work will probably be done by word of mouth.

"How will customers take advantage of processor time?"
Depends on the type of customer, but basically they upload the data to the network to get processed, then they get it back finished.

"How will you manage multiple project requests for processor time?"
There are many ways this could work.
A. Split up the computers into separate cluster groups.
B. Share the time, one pack one type of data the next back another type.
C. Simply go by which project was first or has the highest level of priority.

Actual render farms, real ones, not the pretend ones in your head split a movie into frames, each node in the farm processes a single frame and returns it. A faster node means a faster turn around time per frame.
If you have two nodes you can do twice the work, not MORE as you suggest. What you suggest is plain illogical.

The reason SETI@home or Folding@home work is because the data transfer is minimal. The output and input data is tiny. With rendering, the most marketable service on this product as I see it the output data is tremendous and would flood a home users internet connection. In fact it would probably take longer to send back a completed frame of animation then it would take to render the frame!
Also it would infringe on most ISPs terms of service agreement as you are using the product commercially.

These projects work for voluntare projects, especially science research because the data input and output is minimal. And people give there CPU time away for this. It may have life as a product for commercial science, but how big is this market?

"Actual render farms, real ones, not the pretend ones in your head"
I have set up a local network farm. Don't get me wrong fast nodes are great, but what matters more is the number of nodes working.

"Also it would infringe on most ISPs terms of service agreement as you are using the product commercially."
No, because the server is on our side. The only ISP terms that I have seen that would get you in trouble are trying to run a commercial server on your home connection. Example: If you where running a Web hosting Server, file sharing Server, proxy Server, etc...

This service would not get you in trouble with your ISP.

"The reason SETI@home or Folding@home work is because the data transfer is minimal."
There are other online servers with a network like this that do render work for 3d Animation. The difference is they don't make any money from it. This render work can be worked on like this.

So, anyway on a positive note:
http://www.renderman.../docs/Rmantools.html
Check this for a few free renderer's in development.

I also researched Elephants Dream, a short film rendered using Blender. This could have been done using done using burp which is the free version of what u propose. They use 200 G5 processors for 125 days to do the rendering. This is worth roughly $120000 in amazon terms(minus upload and download traffic) and it was only a 12 minute clip. If you could maintain a network of 10000(being home computers more would be needed!) computers @ 50% availability you could pull $1200 per day if you kept it full of jobs.

The obstacles being porting a renderer capable of commercial films to run on Boinc. It has to be compatible with the big modeling packages such as Maya, light wave and 3d studio. Also security, Peter Jackson wouldn't be happy with stray frames flying across the internet before its released.

This idea should maybe be re scoped as: Building a commercial quality film renderer to run on top of Boinc and founding a company to market and support a rendering network using the Boinc framework and commission jobs for said network.

even tho you have to pay for their services, I think Amazon's EC2 and S3 services could be a competitor since they are a well known company.

Wish your supercomputer services will reach Philippines soon, Kevin. Keep it moving. Good luck!

Here is an example of how someone is utilizing a render farm ... an open source, fractal generating screensaver that uses muliple cpu's to render the ever evolving design.

http://electricsheep.org/tour/

perhaps emulating what others have done will help move it forward.

Yes, there would be a need to port commercial packages to our system. Which is not that extremely hard, considering all major commercial packages can use external rendering programs. I don't know if we would be running on the existing Bionic framework. Though that might be a possibility.

I think Amazon's EC2 and S3 would be a competitor, but they have there pros and cons as well.

Yes, there are other networks that have completed super computer networks. That was discussed and mentioned several times. Thank you for the link.

Anyways, it looks like this idea has won! With a total score of 148!

I like it. It would be good for animations. Every computer in the network would have to be able to interpret the 3D mesh data exactly the same and be able to render it the same way though.
You could use standard plugins. Upload your animation file to the server if it is supported, or one frame at a time. The server can split it into individual frames for you if it did get the animation file (this works for low bandwidth internet connections).
The frame files can be a standard format that every computer can read.
It can then send you a list of the most popular renderers that are online, and what specifications they have so you can choose the most suitable. (you may have a thousand frame animatic and only 10 computers have top quality renderers, but 500 have simple ones).
Then indicate whether you want individual frames (select compression) to put the video together yourself or select a codec for the server to send you a complete video (this works for low bandwidth connections).
The more work you get the server to do the more points it will cost you, and the more frames you render the more points you get. The server can occasionally send you pings and latency tests to determine how many frames to send you or what kind of a discount you'd get for asking it to do a little more work (see above).

This Idea rocks!!

Well, I as for the exact payment system. That will probably get worked out more in depth after everything is completed and after a few tests.

Most likely all projects will use as many computers as possible that can work on the project. That way the work is completed faster.

Anyways, the main thing here is FUNDING and WORKERS. If you would like to fund or work on this project private message me today.

I like it.

Great idea Kevin!

The difference I see in my computer when it is idle vs working on a hard problem is idle temp: 95F, working temp: 120F - this with constant fan speeds. 120F isn't so bad that I need higher fan speeds (the Athlon is pretty efficient even at full load), but I see and feel that increase in air temp, and it's also a cost on my electric bill of a few cents a day (maybe 60W, or 1.5 kWh/day - say $0.20/day, $6/month, or over $70 a year - not counting any increase in the cost of cooling the house in the summer.)

So, the CPU time isn't really free to me - cheap maybe, but the cost is measurable.

Other machines like my laptop start blowing their fans when working on a hard problem... annoying, even if you don't mind the cost.

You math is a bit off, for cooling a single computer.

The average computer (The entire computer not just processor) emits 600 BTU/hr to 720 BTU/hr running full load. At 10 cents per unit of electricity it costs about 6 cents to cool 5200 BTU. So, your computer running max load will cost you about $0.0083 USD to cool. Less then a penny an hour!

Also, this is assuming that there is no normal heat to air exchange. I find it silly to argue cooling costs unless you are setting up a server room with lots and lots of CPUs in a small confined area. This is why computer BTUs rates are only listed on servers. Because the cost is extremely small for home users.

ughhh..very unoriginal!

I have a feeling this will be another me too product looking for crumbs on the target market and probably wont give a better service due to lack of r&d to reckon with as to deliver the feature.

am sorry to burst your bubble your paint 2.0 is better!

-Michael

PS. I agree with Patmania in most aspect.

Here is the developing site link:
http://www.openrendernetwork.com

Also, just a note I can't be on all the time to reply to comments because of my current situation. So, if I don't reply quickly it is not because I don't care.

"I have a feeling this will be another me too product looking for crumbs on the target market"
This is a very new market it is not just bread crumbs. Even amazons local cloud project is very new, the thing is still beta.

"Can you give me examples on what benefits I would have if I joined the network as a personal user. Running the client on my computer.

Right now I am running the same one that comes with the PS3, it helps research on rare deceases."

For personal use so far here is how it is set up.

If there is for profit work available:
You computer will earn points based on the work done (You have the option to set CPU cycles and all of the technical details if you wish).
Points can be used for render time (Use the network yourself) or they can be redeemed for Money/Rewards.

If there is no for profit work then that time will be given to non-profits: Similar to Google AdSense when there is not paid work they use PSAs.

This means if your computer is not giving you points, you will be helping solve scientific and medical problems around the world.

You have the option to enable/disable non-profit work mode. Also, the option to set what work you prefer to do. Example: If you would like to research on rare deceases over climate research.

It does not matter party you are from when it comes to Open Render.

Liberal with the profits, Conservative with start-up costs - Vote for Open Render today!

Also, one more thing I have just started setting up a physical hardware a multi-processor Xeon server to help start up work on the network client program side of things.

Next, SHOUT OUT to zenanthor for getting the web site portion hosted and online.

Nice concept, certainly an extension of distributed computing platforms like Boinc, and I think this technology is seeing steady growth, but it raises some big questions in my mind. How do you compensate the users? How do you monitor cycles, or in other words, what is your common factor by which a user would accrue payment? Would you have to develop a proprietary technology to split any type of file, or just certain ones, while preserving data integrity? How would the actual rendering software handle file segments as opposed to traditional whole files? What about licensing to process proprietary file formats? Would you only support open source rendering tools?

"How do you compensate the users?"
Rewards points: That can be redeemed for use on the network or cash or prizes.

"How do you monitor cycles, or in other words, what is your common factor by which a user would accrue payment?"
The same way every network measures things by the amount of packets processed.

"Would you have to develop a proprietary technology to split any type of file, or just certain ones, while preserving data integrity?"
Possibly if this was needed. It depends on the type of data being computed.

"How would the actual rendering software handle file segments as opposed to traditional whole files?"
Rendering software can all ready do split-frame rendering instead of single frame rendering. It just depends on the type of file you are trying to render. Split-frame rendering (splitting a single frame into a grid) is usually used for large format or high-end stills. For animations usually each node will render by single frame (splitting the entire file into frames).

"What about licensing to process proprietary file formats?"
Most major 3d modeling programs encourage development for the rendering side of things. The only licensing issues would be with other 3rd party rendering programs. I am still looking into the costs but they should be similar to what other rendering farms pay.

"Would you only support open source rendering tools?"
The plan is to add support for all major programs. I was just talking about testing the system on something open source first because you can work out the code faster.

Hey man great Idea,

Im just wondering if the in and out flow of information will just bog down your network/ cpu? It reminds me of Folding@Home.

Keep it up,
Pat

Oh and you'd think Pat mania wouldnt be such an asshole with all that "helpful" info he's giving...

Kevin,
one way to think of this concept in the form in which you've proposed it is like a giant "capacitor" for one's computing cycles, a portion of which are "paid" to the central entity which then resells them for profit. Similar to how solar panels charge up a capacitor over a long period so that the user can draw down quickly on its power, this is the equivalent: you contribute cycles over time and slowly build up your "bank" of CPU's which can then be immediately converted to massive/fast processing power on demand.

I do have the same concern as others as far as the realism of building this within CH. You're looking at some heavy-duty development to pull it off and from what I gather the experience of most people in CH is more in building web apps. One way to side-step this giant dev hurdle might be to leverage xgrid for Macs-> http://www.apple.com...cosx/features/xgrid/
essentially build a portal site that allows people to sign up to participate in a private xgrid and contribute & consume cycles via VPN. Not sure logistically how this would work exactly but if that cpu-sharing functionality already exists, you definitely don't want to expend energy reinventing it.

The similarity advocatusdiaboli mentioned w/ my Render Juice idea I had submitted long ago is actually complimentary in nature. Render Juice could potentially swap out Open Render Network for EC2. RenderJuice is a virtualized rendering engine with hooks for the major 3d design applications - it is agnostic of the utility compute fabric on which it sits.

sean

Interesting analogy.

"realism of building this within CH."
"You're looking at some heavy-duty development to pull it off and from what I gather the experience of most people in CH is more in building web apps."
I think there are some programmers on CH and it is not exactly hard to get some. Also, it is not all client program setup there is the web side of things.

Over all the main thing is funding it takes money to get things done.

Anyways, I hope everyone votes for Open Render. =)

Kevin_Cox: Your idea is a good one, but some important aspects are a little vague. Where do you plan to find programmers for ORN? I ask because as far as I know, there is no one on CH who specializes in the work you are asking for; network programming, an expensive and rare find. In addition to this, you would want to have at least a small team of these programmers to complete ORN in a reasonable time - we are talking years, not months.

In order to make ORN a reality it will cost you a lot more than what CH is giving to the ToC (Tournament of Champions) winner. The ToC money should be considered a secondary form of funding for an idea such as yours. Do you have a plan for primary funding? If so, have you already taken steps towards receiving it?

Now I am not trying to discourage you at all, because your idea is a good one - I am just making sure you are fully aware of what you are getting yourself in to. If you win the ToC money, the community expects to see steady progress on the idea.

Where do you plan to find programmers for ORN?
Online project postings, I all ready know a few that are interested.

"we are talking years, not months."
Well I never said anything about time-lines of months. I don't know about years, maybe a little over a year to make a working client and another year for the user interface website. Over all the networking concept is not that new. I know a that a few people have developed similar non-profit networks with only a very small team.

Also, currently in ideawars just about all of these ideas are going to take at a year or more before they can actually work. Rome was not built in a day.

"Do you have a plan for primary funding? If so, have you already taken steps towards receiving it?"
Over all, ToC money would be a good starting block. Because you need to show some actual development before you are going to get any hard core investors. Currently there are a individual investors willing to put forth moderate amounts and I have a number of people that have all ready contacted me about investing.

As far as the amount needed for this to fully work it is actually not that much. Because there is a very minimal investment in hardware. So, compared to starting a render farm with the hardware it would be a cake walk.

Good answers! Now that you've answered those questions, I feel more comfortable voting for your idea. I have a few more questions, though, if you don't mind.

You said you've found a few people, from online project postings, that are willing to work on your project. If you project development time for the client to last longer than a year and you are only hiring programmers for that set amount of time, what do you plan to do after the client's release? What I'm getting at is patches and bug fixes. Do you plan on having a team of developers whose job is working on ORN or do you plan on contracting work when ORN needs updates?

You have a very good idea, that is why I am asking you all the tough questions. =]

Kevin- have you looked into riding this ontop of xgrid? The Mac community is admittedly smaller than the PC community but it's also much more vocal and self-referencing. There's plenty Mac users to make this service compelling and if it becomes popular you can afford to hit the PC market. As Fitzpatrick says, the programming challenge of recreating the peer-to-peer CPU-sharing functionality is massive and those programmers will command much higher salaries than web developers. You'll need an extremely competent technical lead in-house.

By leveraging xgrid you can focus your dev efforts on simply making a desktop client for VPN'ing in users to a massive distributed xgrid. And even that client app could wrap something like OpenVPN or Hamachi with a custom UI on it so development is minimalized. Unless you have investors that are willing to take a blind leap on this, it seems the key is getting something functional (however small) in place to prove the concept. You'll also get a better valuation from investors and be in a stronger negotiating position if you have beta testers or customers before courting money. I would give a long look at xgrid before starting down the path of developing this app from scratch.

Sean

"What I'm getting at is patches and bug fixes. Do you plan on having a team of developers whose job is working on ORN or do you plan on contracting work when ORN needs updates?"
There will most likely be a mix of in-house and freelance developers. Also, a prior beta test to check for errors before any for profit work gets started. Most work will try to be done in-house unless otherwise needed.

"peer-to-peer" I am thinking more along the lines of server based. But, it could end up somewhere in between.

"have you looked into riding this ontop of xgrid?" I have not had the time to look into the mac xgrid fully yet. But, I assure you that I will look into it thanks for the help.

Capitalization of this project will be quite large and this project will take a long time. Do you have the management experience to pull off such a project. What type of management experience do you have?

Because your project will take a long time to implement, what is to stop a direct competitor with a lot of capital from doing this same thing?

"Capitalization of this project will be quite large and this project will take a long time."
The investment amount is not that big and the time is not absurdly long. It is quite reasonable in comparison to all of the other ideas as far as time.

"what is to stop a direct competitor with a lot of capital from doing this same thing?"
Well it is a relatively new market, very new. We could get in the door with little to no competition. But, I guess whats to say Bill Gates can't rule the world.

Distributing on standard machines a highly-parallelizable task is already done. BOINC, Xgrid, Electric Sheep, distributed.net (many of them already cited here) come to mind.

AFAIK none is generic and reasonably open/free, therefore I submit http://www.cpushare.com/