2024-05-16 09:13 AEST

View Issue Details Jump to Notes ]
IDProjectCategoryView StatusLast Update
0000070mercuryBugpublic2009-02-19 18:08
Reporterpbone 
Assigned To 
PrioritynormalSeverityminorReproducibilityhave not tried
StatusclosedResolutionfixed 
Product Version 
Target VersionFixed in Version 
Summary0000070: Suspect that deep profiler reports incorrect call sequence counts for recursive predicates.
DescriptionThe call sequence counts reported for recursive predicates don't seem to make sense.
Additional InformationWhen profiling icfp2000 navigate to the eval.interpret/3-0 procedure. See mdprof_cgi1.html Note that the call site eval.interpret/7-0 (a different predicate) has call sequence count: 969,620,690.
Click this call site to move to the eval.interpret/7-0 procedure.
The self call sequence counts are now reported as: 626, Is this correct? Why is it not 969,620,690? See mdprof_cgi2.html
Note that the total call sequence count is equal to the self call sequence count. This is incorrect unless the call has no child calls. The total call sequence counts per call (1.0) is also incorrect.


TagsNo tags attached.
Attached Files

-Relationships
+Relationships

-Notes

~0000108

pbone (administrator)

To see this for yourself: http://plum.csse.unimelb.edu.au/cgi-bin/mdprof_cgi?proc/225&p-_-S-_-_/box/cols/5/nosum/context/nc/th/hh&/home/pbone/mercury/icfp2000/Deep.data

~0000157

pbone (administrator)

This is not a bug, just a misunderstanding.
+Notes

-Issue History
Date Modified Username Field Change
2008-07-30 14:13 pbone New Issue
2008-07-30 14:13 pbone File Added: mdprof_cgi1.html
2008-07-30 14:13 pbone File Added: mdprof_cgi2.html
2008-07-30 14:20 pbone Note Added: 0000107
2008-07-30 14:21 pbone Note Deleted: 0000107
2008-07-30 14:21 pbone Note Added: 0000108
2009-02-19 18:08 pbone Note Added: 0000157
2009-02-19 18:08 pbone Status new => closed
2009-02-19 18:08 pbone Resolution open => fixed
+Issue History