SETI

Hi Philip,

There's no single location of documentation, but there's community-created content here: Raspberry Pi Projects (there are beginners articles and projects there) and beginners journal-like articles at http://www.raspberrypi.org/magpi

The installation of boinc is discussed here: https://www.raspberrypi.org/magpi/share-boinc/

Regarding clusters, unless software is specifically designed to run like this, the software needs to be run individually on each machine, there's no sharing of resources normally. You can log into one device and remotely access the others, using a few different techniques depending on needs.

If you wish to view a full graphical desktop, then tools like rdesktop can be used. See the Remote Desktop section here to see how to do that: Accessing and Controlling the Pi

If the software you're using is purely text-based and has no graphics and if you are happy with just command-line interaction (a bit like DOS) then it's possible to use a protocol called SSH. It is described at that link too.

Just for reference, if you've already got a PC that you use for day-to-day stuff, then if you don't need graphics (or don't need intensive graphics) there's no need to attach a keyboard/monitor/mouse to a Pi, it can be accessed using the remote methods mentioned above, from your PC. My Pi is not connected to display/kb/mouse, just to the home network. I connect from the PC to do some work on it.

I don't use the graphical view, so this is how it looks like from my Windows PC (red is Pi, blue is a different Linux machine.. but the same method can be used for many Pi's). So, for graphical method or not, it mainly depends on the end application and also on what you prefer, and desired mode of use - some prefer a separate keyboard/mouse/display).

Shabaz, maybe I am wrong but as far as I know, this program, as well as many other, ha very few to do with the clustering of a local device. The cluster is outside and every connected device just sends its contribution in data analysis controller by the remote engine, hopefully using the unused CPU time.

Local clustering with the Raspberry PI is possible –always based on my reduced experience in this case – but it is focused on a totally different topic. In the past, I have used a cluster of different computers (including some raspberry PI B) and some netbook with ubuntu to increase the speed of some slow tasks, with very good results; on a cluster of five different devices (with different CPU freq.) the average results have been:

• Rendering 3D objects with 3D Studio MAX and Blender: a global increase of about 20%
• Assembling and final rendering of large video productions and a lot of editing features using Cinelerra: 35% with its own very efficient headless server
• Compiling long sources (rebuilding the entire Linux kernel with custom drivers) with concurrent/distributed compiling (you can find some notes here and there: https://github.com/alicemirror/Ltib-EA3250 ): Probably the most efficient way

Consider that I have always searched for and privileged the methods that share the resources on a server-clients basis accordingly with the different performances of the clustered processors. This means that every contributor device is used at the best of his performances. For example, assembling videos with Cinelerra every client CPU, memory, etc. is tested by the server to decide the number of frames-per-time-unit should be sent to every remote module.

Just to show an example, below a video of a series of 12 for a DVD I produced several years ago; all the pieces (and the entire DVD assembly) were launched in a batch together over multiple computers.

Edit media

Dimensions x Small Medium Large Custom

Subject (required) Brief Description Tags (separated by comma) Video visibility in search results Visible Hidden

Parent content

RE: SETI

Poster

Upload Preview

Ok

Cancel

Shabaz, maybe I am wrong but as far as I know, this program, as well as many other, ha very few to do with the clustering of a local device. The cluster is outside and every connected device just sends its contribution in data analysis controller by the remote engine, hopefully using the unused CPU time.

Local clustering with the Raspberry PI is possible –always based on my reduced experience in this case – but it is focused on a totally different topic. In the past, I have used a cluster of different computers (including some raspberry PI B) and some netbook with ubuntu to increase the speed of some slow tasks, with very good results; on a cluster of five different devices (with different CPU freq.) the average results have been:

• Rendering 3D objects with 3D Studio MAX and Blender: a global increase of about 20%
• Assembling and final rendering of large video productions and a lot of editing features using Cinelerra: 35% with its own very efficient headless server
• Compiling long sources (rebuilding the entire Linux kernel with custom drivers) with concurrent/distributed compiling (you can find some notes here and there: https://github.com/alicemirror/Ltib-EA3250 ): Probably the most efficient way

Consider that I have always searched for and privileged the methods that share the resources on a server-clients basis accordingly with the different performances of the clustered processors. This means that every contributor device is used at the best of his performances. For example, assembling videos with Cinelerra every client CPU, memory, etc. is tested by the server to decide the number of frames-per-time-unit should be sent to every remote module.

Just to show an example, below a video of a series of 12 for a DVD I produced several years ago; all the pieces (and the entire DVD assembly) were launched in a batch together over multiple computers.

Edit media

Dimensions x Small Medium Large Custom

Subject (required) Brief Description Tags (separated by comma) Video visibility in search results Visible Hidden

Parent content

RE: SETI

Poster

Upload Preview

Ok

Cancel