1.1.4. Parallelization efforts for Molcas modules

Presented below is a table of modules in Molcas that can benifit from parallel execution through distribution of work and/or resources. If a module is not listed in this table, and the module-specific documentation does not mention anything about parallelization, then you have to assume the module is not (efficiently) parallelized. This means that even though it will get executed in parallel, all processes will perform the same serial calculation! Be aware that for parallel modules with serial components, the use of the serial components (indirectly or through the use of a keyword) might adversely affect CPU and memory usage for large calculations. In that case, you might have to increase the runtime or memory, or avoid/use keywords that activate/deactivate the serial components.

Table 1.1.4.1 Modules in Molcas which benefit from parallel processing.
Module Parallel speed-up expected for Notable non-parallel parts
SEWARD
conventional 2-el integrals
Cholesky vectors
1-el integrals
Douglas–Kroll–Hess
properties
SCF
orbital optimization
properties
RASSCF
orbital optimization
CI optimization
properties
MBPT2    
CASPT2
Cholesky vectors
conventional 2-el integrals
properties
multi-state interaction
ALASKA
displacements (if using numerical gradients)
 
GEO
displacements