From: Peter Z. <pe...@in...> - 2010-05-06 20:41:49
|
On Thu, 2010-05-06 at 11:33 -0500, Maynard Johnson wrote: > > 1. perf does support native events, but right now, it's not user friendly -- the > user needs to know the appropriate hex code value that corresponds to each > native event. Since you need the arch docs to find and make sense of them typing values in hex doesn't seem like the threshold. > 2. perf-report does not output XML. There are multiple GUI tools available that > make use of oprofile XML output. XML the only format impossible to parse by both computers and humans! > 3. perf has no mechanism for archiving sample data and binaries for off-line > analysis. Oddly enough its called: perf archive > 4. perf has no corollary to oprofile's opimport utility -- useful for offline > analysis on a different architecture from the profiled system (typically used > with embedded systems). Works just fine. But we don't need to import stuff, we can simply use perf report on the the remote host after you've untar'ed the perf archive result. > 5. perf has no capability for profiling JITed code (Java is the only language > officially supported by oprofile right now). We've had Jato working since the very early days of pref. Granted its a tad rough, but could easily be extended into a library to be used by other JITs. > 6. perf doesn't have a "diff" feature to show side-by-side differences between > two profile runs. perf diff work very nicely |