Sign in
Microsoft.com
 
Microsoft
 
 
 

none
Pooling Several PCs together to run CPU intensive apps

 > 

    Question

  • Question
    Sign in to vote
    0
    Sign in to vote

    I am trying to dtermine the most cost effective why to get the CPU horsepower I need for my music studio work.

    For music studios they make expansion cards that help onload the demand on the CPU as the CPU has limits.  As CPUs get faster the move is to not use such cards and go Native (allowing the CPU to do all the work).

    I am finding for big projects Native still has its limits.  I am curious if there are ways to pool several PCs together to share the load.  Do a a web search I found Server 2008 HPC - it seems like it is this concept but to distribute the computation load.  Either the apps need to be designed to allow this or some "programming" needs to be done to accomplish it.

    Am I correct on Server 2008 HPC - and are there any other options that anyone on this forum may be able to recommend?

    Dave

     

     

     

    Tuesday, January 18, 2011 2:17 AM
    |

All replies

  • Question
    Sign in to vote
    0
    Sign in to vote

    Windows HPC Server / Microsoft HPC Pack-based clusters are probably not the solution to your problem. As you've read, an HPC Pack-based cluster is able to accelerate certain types of applications that either work independently to take an input->create an output or that have been written to work together to accomplish a larger goal. It is not designed to directly help offload CPU usage for single apps.

    I assume by 'expansion cards' you are referring to things like the various MoTU/Apogee/Avid/etc DSP and I/O cards. Those need really high bandwidth to do their job, so trying to move some of that off the box would likely just slow things down ;)

    A cluster would be more helpful for tasks like normalizing thousands of samples or compressing your masters in to multiple different bitrates for distribution on the web.

    Friday, February 4, 2011 8:36 PM
    |
    Moderator