1#!F-adobe-helvetica-medium-r-normal--18*
2#!N
3#!CNavyBlue #!N  #!Rall272
4Parallelism for Data Explorer SMP #!N #!EC #!N #!N For completeness,
5the notion of module parallelism is discussed here. If you are
6developing visualizations or modules exclusively for use with the IBM Visualization
7Data Explorer running on a single-processor workstation, then these concepts are
8not applicable. However, if your visualizations or modules are to be
9run on both the IBM Visualization Data Explorer and IBM Visualization
10Data Explorer SMP, then these concepts are important for achieving higher
11performance. #!N #!N Every module that performs any significant amount of
12processing is "parallelized"; that is, the module makes use of all
13processors made available to Data Explorer to operate on the data.
14#!N #!N Data Explorer uses explicit data partitioning as the primary
15framework for parallelism. Data Explorer partitions the data into local, self-contained
16regions. In general, visualization modules then generate subtasks corresponding to partitions.
17For more information about partitioning, see Partition in IBM Visualization Data
18Explorer User's Reference. #!N #!N In general, parallel programming is complex.
19To help manage it, Data Explorer simplifies the process by providing
20a simple fork-join parallelism model to implement coarse-grain shared memory parallelization
21(data parallel). Using data partitions, read-only objects, and a single-fork join
22mode simplifies the module writing task by avoiding the explicit use
23of locks in modules, thereby reducing the possibility of deadlock. For
24information about adding modules to the Data Explorer system, see IBM
25Visualization Data Explorer Programmer's Reference. #!N #!N #!N  #!F-adobe-times-medium-i-normal--18*   Next Topic
26#!EF #!N #!N  #!Lunduse,dxall274 h Graphical User Interface: Basics  #!EL  #!N  #!F-adobe-times-medium-i-normal--18*   #!N
27