An Evaluation of High-Performance Functional Computing


Project Leaders


Objectives


An example using a FORTRAN molecular dynamics code

In the original code, 90% of the work was done by 10% of the code (the modules NON-BOND and OTHER-FORCES). This kernel was parallelized by converting it to SISAL. This had some impact on FORCE-RUN and CALC-FORCES, but the rest of the code was unchanged. This leveraged the development time put into the old code (I/O interface, etc...) into the new parallel code. Parts of the algorithm were rethought, but the effects were confined.


Mixed-Language Debugging


Distributed-memory Implementations


Impact

Low-cost development of parallel portable software through:

Current Work


Back to PDPC Projects