Talk:ITK Working Group Parallel Computation: Difference between revisions
Kevin Hobbs (talk | contribs) (Big Memory) |
Kevin Hobbs (talk | contribs) (Stream -> MPI) |
||
| Line 8: | Line 8: | ||
How could a cluster be used to allow ITK to handle Images larger than the memory on a single machine? | How could a cluster be used to allow ITK to handle Images larger than the memory on a single machine? | ||
== Stream -> MPI == | |||
I'm trying to figure out how hard it would be to go from a streamable filter to a distributed memory filter. | |||
Latest revision as of 21:10, 18 January 2005
I read here that there is no support for clusters and ITK. This is a shame, but there must be a way to work around it. I'm thinking of loading different Regions Of Interest on each node. If whatever algorithm has edge effects, the ROIs might overlap just enough to eliminate them.
Build with mpiCC
I used mpiCC for a build on our cluster and I get a lot of segfaults. I can do examples like ImageReadRegionOfInterestWrite, RGBToGrayscale, and RGBImageSeriesReadWrite but MedianImageFilter and WatershedSegmentation1 segfault. I assume they use more arithmetic. I did "make test" and saw a failure with VXL fly by. How should I investigate?
Big Memory
How could a cluster be used to allow ITK to handle Images larger than the memory on a single machine?
Stream -> MPI
I'm trying to figure out how hard it would be to go from a streamable filter to a distributed memory filter.
