|
|
Use Matlab with Xgrid
Wow... I'm using matlab for my project to do LOTs of fft & ifft in a loop, this might really make things much faster. What kind of speed up did you manage to get? I would imaging Xgrid will have some overhead associated with splitting the work load on multiple machines.
Use Matlab with Xgrid
This hint isn't really for doing any fine-grained parallelism, its more for "embarassingly parallel" tasks, and you pretty much have to split up the work by hand. In the example I gave, matlab will launch and immediately run foo.m on each machine, and the input argument will be a number unique to that instance of matlab which you can use as an index into "work chunks". Then you would write code in foo.m to decide which chunk of data to work on based on that index. In this kind of situation speed scales linearly with processors, minus whatever overhead there is passing around your data.
I have done some more fine-grained parallelism with matlab based on the "MatlabMPI" toolbox. This doesn't use Xgrid or have any sense of load balancing however. I wrote my own wrappers around this to do some ad hoc load balancing, but this is basically in-house research code, not ready for prime time yet. In the meantime, The MathWorks just released a new toolbox for distributed computing. I haven't tried it yet, but it looks quite handy. -Ian
Use Matlab with Xgrid
There is a free webinar on December 9, 2004 for the Distributed Computing toolbox: |
SearchFrom our Sponsor...Latest Mountain Lion HintsWhat's New:Hints1 new Hints in the last 24 hoursComments last 2 daysNo new commentsLinks last 2 weeksNo recent new linksWhat's New in the Forums?
Hints by TopicNews from Macworld
From Our Sponsors |
|
Copyright © 2014 IDG Consumer & SMB (Privacy Policy) Contact Us All trademarks and copyrights on this page are owned by their respective owners. |
Visit other IDG sites: |
|
|
|
Created this page in 0.13 seconds |
|