Take the 2-minute tour ×
Super User is a question and answer site for computer enthusiasts and power users. It's 100% free, no registration required.

I have 3 pc's and was wondering if i can connect them into a cluster. I read few articles about clustering, but there is one think which I'm not sure about. If I build my cluster can I run apps for "normal" os on it(i don't know, for example my eclipse IDE:D) and if they wiil run more efficient? I think clusters are rather use when it comes to servers, storage...

Maybe the answer is kind of obvious, easy but I'm not specialist in this sector:D

share|improve this question

migrated from stackoverflow.com Jun 20 '11 at 18:55

This question came from our site for professional and enthusiast programmers.

add comment

2 Answers

Clustering will (usually) have no positive effect for interactive programs, since they can rarely be parallelized. Eclipse is heavily parallelized, but I'm afraid you will not see a significant speedup for day-to-day actions like searching, switching tabs, etc.

However, compilation greatly benefits from parallelization (and therefore executing in a cluster). For example, if you're compiling C/C++, you may use distcc. A few other actions can also be sped up in a cluster, like 3D and movie rendering, automated tests, and scientific calculations.

share|improve this answer
add comment

To run applications in parallel on a cluster, they have to be specifically enabled to do so. An application that is capable of running on a multi-core architecture is not necessary capable of using multiple cores spread across a cluster, either. This is "Shared memory / threading" (multicore) versus "distributed memory / message passing" (cluster) distinctions in the world of parallel programming.

If you're using Linux, there's plenty of free software to help enable cluster computing. Start at http://www.beowulf.org for some pointers. If an application says it was built with MPI, then it's ready to go on a cluster. Other tools that say they support distributed computing (e.g. gmake -p) should also work with a bit of setup.

If you're a windows user, then Windows HPC Server 2008 R2 is the way to go, as it will have the software stack you need to begin exploring - but it ain't free!

If you're interested in learning about cluster computing, then building your own is a great way to go, and will give you some bullets on your resume that look good to employers. If you're just looking to slap something together quickly to get more performance out of an existing set of applications, you'll likely be disappointed and frustrated.

Good luck!

share|improve this answer
add comment

Your Answer

 
discard

By posting your answer, you agree to the privacy policy and terms of service.