[no subject]

[chang@cme.nist.gov (Forrest Chang)]

Subject:  Paging and performance.


	Has anyone running CLUE on a symbolics suffered a paging/performance
problem between applications written in clue 1-13 and clue 1-15.  We are
metering the application to try to narrow down the culprit.  Unfortunately,
each metering run takes about 2 hours so we have only 4 sets of data in which 
there are two pairs of data that are similar, but unlike the other pair.
	
	Alot of thrashing seems to concetrated in calls that handle
events.  The application in quesiton is the same tree editor that has
be mentioned before, the performance problem occurs when we try to
draw the tree (i.e. making and mapping all these contacts as nodes).

	Apparently (which I've found is often very misleading) make and map
lots of contacts == thrashing about.  Sound familiar to anyone?

	In addition I will be bringing up clue 6.0 sometime soon, maybe
that will help.


	Thanx for any help

	Forrest

	NIST