I think that it's not possible (maybe there is a way to do it... I don't know).


If you use BOINCTasks or BOINCView (other very good tool option), in both cases the management could be made by one centralizer computer, but the downloads/uploads will be made by each machine (under your commands using the centralizer one, or each machine alone following your WCG profile web settings configurated).

For you and for ericinBoston, I think that could be a plus read the follow thread, to get additional information about how to install the Boinc client in many machines at sametime:

http://www.worldcommunitygrid.org/forums/wcg/viewthread_thread,28514

Good look !

Ericinboston:
While I admire you sprit and dedication, this is an ill-conceived idea. Before you even start planning how you are going to build, monitor and manage such a network, you need to plan (and resolve) the huge problems you will have with power and heat management. An “unfinished basement” is unlikely to have any where near the electric circuits you will need for 100 machines. Once you have calculated your power load, you need to contact a licensed electrician to design a safe power layout. Also, unless you want to burn out some percentage of your power supplies every time you have a power spike, you will need a rather sophisticated UPS system to provide clean power and backup. Large-scale UPS systems are quite expensive and can be temperamental. .

You need to calculate your heat load. Opening windows and pointing vents at machines is very unlikely to work. Believe me I know that from personal experience with my farm. Six hundred square feet is not a large space unless you have a very high ceiling. You need to calculate the cubic feet of air rather than just square footage. Again you need to consult an HVAC professional to determine your needs.

Now you can think about how you are going built and manage such a large system/network. You have some serious problems to resolve there too as Ker and others have pointed out.

Personally, I am certain you would get much more “crunching” for your time and money if you were to built a few muli-processor, multi-core, milti-card machines which would greatly reduce the “overhead” required to support 100 machines. However, in the end, it is your time and your money. Good luck!

Cheers!



PS: Thanks KerSamson for you comments!

Interesting perspective and point of view, but for researches and projects, were frequently we need total control, ilimited acess, and mainly 24x7 time full dedicated, ever with frontier conditions around the expriments and tests under total supervisioning, I have doubt that it could be made in this way.

But it is an inspiring idea !

In that case, I gotta wonder why Mac Mini in the first place. Is it an OS preference? A size issue? Power consumption? (have you considered Atom processors if this is the case) Bang for your buck, you really can't beat multi-core/thread with multiple CPU's in a single box. Not to mention throwing in systems that can run up to 4 GPU's at once for DC'ing other projects that support GPU crunching. Who knows...one day maybe even IBM or WCG will have an ap for that!....lol (pardon the Verizon joke)

It is if you are only running let's say 15 systems instead of 100 for the same or more total threads.

Also, maybe you should look into this. You never know...you might get lucky.

http://blogs.amd.com/work/2010/03/03/48-cores-contest/

Edit: Also keep in mind that this gives you the option of up to 4 GPU's and not necessarily using all 4 of them to begin with. The Mac Mini doesn't have the option.