On the 0x24C day of Apache Harmony Geir Magnusson, Jr. wrote: > On Dec 27, 2006, at 2:52 AM, Egor Pasko wrote: > > > On the 0x214 day of Apache Harmony Geir Magnusson, Jr. wrote: > >> Is this tuff documented? Wanna throw it in the wiki or a patch for > >> the site? > > > > (...raising the thread from the ashes...) > > BTW, recently Nikolay Sidelnikov wrote the draft version of the doc as > > a standalone page. Interested parties can look into HARMONY-2865 > > > > Armand, you may find it extremely useful :) > > > > Nadya, Sveta, could you, please, look into the doc?.. > > Nice - it would nice also if Nikolay had the time, to augment that > with an example output of the .stat file, so people using this for > the first time can feel comfortable that they got the right thing, > and maybe some notes on how to read the .stat file.
FYI: Nikolay has updated HARMONY-2865 with a sample .stat file. Nadya, how can I help to put the doc on the website? > geir > > > > >> Egor Pasko wrote: > >>> On the 0x214 day of Apache Harmony Mikhail Fursov wrote: > >>>> + Some usage hints: > >>>> 1. You can get any stylesheet editor (like Excel), open iprof > >>>> data and build > >>>> graphs from the colums - and you will have very useful pictures. > >>> gnuplot :) > >>> > >>>> 2. The disadvantage of iprof is that it counts blocks, insts and > >>>> helpers > >>>> calls and does not count time spent in helpers and blocks. You > >>>> still need > >>>> VTune to get this info. > >>> 1 more disadvantage: counters are not synchronized resulting in > >>> somewhat inaccurate data for multithreaded apps > >>> > >> > > > > -- > > Egor Pasko > > > > -- Egor Pasko
