[Also, related, why do people still introduce new tools that have generic, and therefore un-searchable names? Google is useless for "oprofile trace difference"...]
I don't think these naming conventions are problematic. Generic and useful tools should be named with simple, generic names. This is Google's problem, or a problem of the programmer using the wrong methods to search for the tool.
It looks like oprofile takes an approach similar to gprof: It tries to show you how much time is spent in each function.
The new trace command, on the other hand, traces kernel events: page faults, system calls, fork/exec commands, and so on. Obviously, there's some overlap between the two activities (particularly as both tools mature), but for now, it looks like the 'trace' command records rather different data than oprofile. Of course, I don't know anything about oprofile beyond what's on their introductory web pages, so maybe I'm overlooking something.
6 comments
[ 3.0 ms ] story [ 27.0 ms ] thread[Also, related, why do people still introduce new tools that have generic, and therefore un-searchable names? Google is useless for "oprofile trace difference"...]
LWN has good comments on the topic: http://lwn.net/Articles/415728/#Comments.
The new trace command, on the other hand, traces kernel events: page faults, system calls, fork/exec commands, and so on. Obviously, there's some overlap between the two activities (particularly as both tools mature), but for now, it looks like the 'trace' command records rather different data than oprofile. Of course, I don't know anything about oprofile beyond what's on their introductory web pages, so maybe I'm overlooking something.
This new "trace" looks like it more about system calls, and perhaps has some overlap with perf (and uses perf).