|
From: Anthony H. <ap...@em...> - 2015-07-09 19:10:40
|
Hi, When running with valgrind massif with --time-unit=B option, how can the time(B) column value be so much larger than the total(B) column value? In this case at snapshot 35, I have a total of 119,461,400 but the time(B) is at more than 1GB. What does that mean? and where does it come from? Thanks, Anthony -------------------------------------------------------------------------------- n time(B) total(B) useful-heap(B) extra-heap(B) stacks(B) -------------------------------------------------------------------------------- 33 1,056,768,488 71,575,048 71,518,410 56,638 0 34 1,083,961,928 98,768,488 98,709,838 58,650 0 35 1,105,439,944 119,461,400 119,397,314 64,086 0 |
|
From: Milian W. <ma...@mi...> - 2015-07-10 09:19:45
Attachments:
signature.asc
|
On Thursday, July 09, 2015 12:10:14 PM Anthony Haas wrote: > Hi, > > When running with valgrind massif with --time-unit=B option, how can the > time(B) column value be so much larger than the total(B) column value? > In this case at snapshot 35, I have a total of 119,461,400 but the > time(B) is at more than 1GB. What does that mean? and where does it come > from? The time(B) just counts how much memory was requested in total, ignoring deallocations, and uses that for a "time" axis. The total(B) counts how much memory is currently being used, i.e. actually takes deallocations into account. HTH -- Milian Wolff ma...@mi... http://milianw.de |