|
Parallel Processing on Distributed Linux Systems Examples |
|
|
Customized Parallelization of Existing Programs |
|
|
|
Projects have been undertaken to convert serial (single CPU) programs into parallel programs to run on a distributed network of machines. Examples include tomographic analysis and seismic migration. The conversions involved the analysis of I/O patterns in order to avoid bottlenecks and allow the application to scale up efficiently when run on multiple processors. |
|
|
|
Platforms: |
SGI Origin, Linux Beowulf Clusters |
|
|
|
Languages and tools: |
C, C++, Fortran, MPI, PVM |
|
|
|
Seismic Reflection Migration on Distributed Systems |
|
|
|
Developed parallelized programs to migrate seismic data, capable of handling data sets of 100 GB or more. The applications were developed for both shared memory machines and distributed processors (Beowulf clusters). Jobs are scalable to 100 machines or more. |
|
|
|
Platforms: |
SGI Origin, Linux Beowulf Clusters |
|
|
|
Languages and tools: |
C, MPI, PVM |
|
|
|