Changes between Version 17 and Version 18 of CloudySummitBrussels


Ignore:
Timestamp:
2009-01-07T08:22:54Z (12 years ago)
Author:
rjrw
Comment:

Move note subsequently added to Parallelization

Legend:

Unmodified
Added
Removed
Modified
  • CloudySummitBrussels

    v17 v18  
    1212
    1313shared vs distributed memory. The current implementation of OpenMP does not support distributed memory clusters. Future upgrades may.
    14 
    15 [Note also: OpenMP 2.5, as implemented in currently available gcc compiler versions, does not allow the parallelization of loops indexed by iterators, or even unsigned integer types.  http://iwomp.univ-reims.fr/cd/papers/TM06.pdf analyses approaches to get around this limitation.  http://wikis.sun.com/pages/viewpage.action?pageId=57508020 suggests that this will be possible in OpenMP 3.0]
    1614
    1715NUMA systems. OpenMP was never defined with NUMA systems in mind. Threads/CPUs share memory and the physical allocation is determined by which CPU touches the memory first. Keeping this optimal as far as access speed is concerned is going to be a nightmare.