Stanford PhD Defense
M.P. Brown
Computational highlights
•
Model space = one CMP location
•
Vertical stretch
•
Coarse-grained parallelization (MPI)
•
In-core optimization
•
20 CG iterations
à
3 hours
•
4 multiple generators
•
Only first-order multiples
•
16 x 1.3 GHz P3
•
~PSDM