CPU/Mobo sharing two cpu to increase PC performance

[devi691]

guyz,
I have a question ,I dont have any high end PC bt I have 2 dual core(2.5GHz) CPU embaded systems . WITH 1 -1 GB of RAM each.
I need to compress 500 GB of DATA with 7z ultra compression.
CAn i pair two computers to increase cpu performance???
I heard about DISTRIBUTED SYSTEM but can someone help me out to set it up???
please guide me gurus

 

[vishalrao]

is the 500GB data in the form of many files? does the end result have to be a single file or can it be multiple files?

why not just spread the initial data into two separate (shared) folders and simply run 7zip from one CPU on one folder and from the other CPU on the other folder.

you will end up with two ZIPs, finally if you need a single zip, just combine the two zips into a single one in "store" mode (no compression) which just joins them together.

 

[prashanthkp]

Myself having a P4 and a P3 system with different motherboard & RAM's. Whether is it possible to connect them both to make it as one and increase my processing power. If so how can i do it ! pls help me on it ?

 

[net]

OS like cloud linux can be used to utilise multiple systems, but you again need to install OS to utilise such technique.

 

[TEUser2K1]

think it will be very difficult for normal users.

Reminds me of:

Beowulf (computing) - Wikipedia, the free encyclopedia

Beowulf.org: The Beowulf Cluster Site

 

[devi691]

bro , I have around 40 folders with different sizes,but thats not the point, i want to share two different computers not for just file sharing but to performance increase.

 

[dovakhiin]

bro , I have around 40 folders with different sizes,but thats not the point, i want to share two different computers not for just file sharing but to performance increase.

Fair enough...but can tell us what you intend to do with the cluster you want to build?...If its for general computing or gaming its not worth the effort.First of all the peak computation power wont be all that much for the config you mentioned and secondly the power wont scale that well especially for real-time use.

 

[TEUser2K1]

What you may need is a setup for parallel & /distributed computing (where beowulf clustering, etc. comes in to picture) to share processing capabilities - which I presume is a difficult task for normal users to configure.

Also check (note Distributed PPSS) ppss - Project Hosting on Google Code

It would be great if you can share your finding /understanding in this thread, which may help others too...

 

[nitant]

Here is how to do it theoretically

1) Get some Beowulf Linux Build. I recommend parallelknoppix or ClusterKnoppix. Have used ClusterKnoppix personally and it was pretty simple.

2) Set up the networking and get the machines to see each other.

3) Because this uses the openMosix kernel, you dont need to do programming, instead just find a multithreaded compression engine, and use it. OpenMosix, will automatically distribute the load.

4) Hopefully you end up with a nice compressed file.

You might wannna read this -> Re: [Beowulf] Accelerator for data compressing

Oh and believe me, your speed will actually double(considering comps are equal and your tasks are CPU intensive.) I have tested this out for a project using Blender and 4 computers. In the end, it took around 30% of the time, so well worth it.

 

[TEUser2K1]

^^ thats good post nitant, thanks...

 

[princeoo7]

can this techni be used for gaming ???

 

[devi691]

Here is how to do it theoretically

1) Get some Beowulf Linux Build. I recommend parallelknoppix or ClusterKnoppix. Have used ClusterKnoppix personally and it was pretty simple.

2) Set up the networking and get the machines to see each other.

3) Because this uses the openMosix kernel, you dont need to do programming, instead just find a multithreaded compression engine, and use it. OpenMosix, will automatically distribute the load.

4) Hopefully you end up with a nice compressed file.

You might wannna read this -> Re: [Beowulf] Accelerator for data compressing

Oh and believe me, your speed will actually double(considering comps are equal and your tasks are CPU intensive.) I have tested this out for a project using Blender and 4 computers. In the end, it took around 30% of the time, so well worth it.

nice suggestion dude, I started workin on it ,lets see what happens,

I think we are gonna find a new world behind the old door.