Tutorial: Magik Performance

Multi Threading Techniques

Gathering Things from Threads

The following code allows you to create a rope of data from individual threads. It uses the atomic_queue to wait for all threads to be completed before returning the complete set.

+ Show Code

_pragma(classify_level=restricted, topic={fcsi,osm})
_global test!fcsi_test_queue<<
    _proc()
    ## Parameters  : 
    ## Returns     : 
    ## Function    : 
    ## Methodology : 
    #---------------- HISTORY ---------------------
    # (11/17/2010 - Mark Field, FCSI): Created.
    #----------------------------------------------
    _global !fcsi_test_queue!;
    !fcsi_test_queue! << atomic_queue.new(6)
    _for i _over 1.upto(6)
    _loop
        _proc()
            test!fcsi_test_queue1()
        _endproc.fork_at(_thisthread.high_background_priority,_self)
    _endloop
    _local retRope << rope.new_for(6)
    _for a _over 1.upto(6)
    _loop
        retRope.add_last(!fcsi_test_queue!.get())
    _endloop
    _return retRope 
_endproc
$
_pragma(classify_level=restricted, topic={fcsi,osm})
_global test!fcsi_test_queue1 <<
    _proc()
    ## Parameters  : 
    ## Returns     : 
    ## Function    : 
    ## Methodology : 
    #---------------- HISTORY ---------------------
    # (11/17/2010 - Mark Field, FCSI): Created.
    #----------------------------------------------
    _global !fcsi_test_queue!
    _local addRope << rope.new_for(5)
    _for a _over 1.upto(5)
    _loop
        addRope.add_last(a)
        _thisthread.sleep(500)
    _endloop
    !fcsi_test_queue!.put(addRope)
_endproc
$