It looks as if the calculation of unaccounted ticks is somewhat wrong I think it should 16 so that unaccounted plus the JavaScript ones sum up to 27/100%. Can you share the new JavaScript code you are profiling?
Regards, Søren On Tue, May 26, 2009 at 14:17, Ondrej Zara <[email protected]> wrote: > Hi Eric and Soren, > > thanks a lot for your suggestions. I tried switching to > linux-tick-processor, which gives more relevant results. > > I am relatively happy with the low tick count (27), but it still puzzles me > why these are categorized as "Unknown". My output now looks like: > > Statistical profiling result from ../v8cgi/v8.log, (27 ticks, 27 > unaccounted, 0 excluded). > > [Unknown]: > ticks total nonlib name > 27 100.0% > > [Shared libraries]: > ticks total nonlib name > > [JavaScript]: > ticks total nonlib name > 2 7.4% 7.4% LazyCompile: <anonymous> > /home/ondras/svn/v8cgi/lib/js.js:0 > 1 3.7% 3.7% Script: native messages.js > 1 3.7% 3.7% Script: native date.js > 1 3.7% 3.7% Script: > /home/ondras/svn/v8cgi/example/wwwclient.js > 1 3.7% 3.7% LazyCompile: shift native array.js:314 > 1 3.7% 3.7% LazyCompile: join native array.js:214 > 1 3.7% 3.7% LazyCompile: SetupArray native array.js:707 > 1 3.7% 3.7% LazyCompile: RegExpConstructor native regexp.js:52 > 1 3.7% 3.7% LazyCompile: <anonymous> native v8natives.js:183 > 1 3.7% 3.7% Function: <anonymous> > > > > Also, I do not fully understand the percentage numbers: which are supposed > to sum up to 100%? The "bottom up" profile part contains basically the same > data as the simple output contained in this e-mail. > > > Thanks, > Ondrej > > > > > 2009/5/26 Søren Gjesse <[email protected]> > > To add to Eriks comment I also suggest that you try the newer JavaScript >> based tick processor instead of the Python based one. Use >> linux-tick-processor instead >> of linux-tick-processor.py. linux-tick-processor is a shell script which >> runs the developer shell (d8) with the JavaScript code to process the log >> file. >> /Søren >> >> >> On Tue, May 26, 2009 at 13:47, Erik Corry <[email protected]> wrote: >> >>> >>> 2009/5/26 ondras <[email protected]>: >>> > >>> > Hi, >>> > >>> > I am a newbie to V8 profiler - I was not able to get it correctly >>> > running. Here is what I do and what I see: >>> > >>> > ond...@kapitan:~/svn/v8$ scons library=shared sample=shell && ./shell >>> > --prof -e "1+1" && tools/linux-tick-processor.py v8.log >>> > scons: Reading SConscript files ... >>> > scons: done reading SConscript files. >>> > scons: Building targets ... >>> > scons: `sample' is up to date. >>> > scons: done building targets. >>> > Statistical profiling result from v8.log, (12 ticks, 12 unaccounted, 0 >>> > excluded). >>> > >>> > [Unknown]: >>> > ticks total nonlib name >>> > 12 100.0% >>> > >>> > [Shared libraries]: >>> > ticks total nonlib name >>> > >>> > [JavaScript]: >>> > ticks total nonlib name >>> > >>> > [C++]: >>> > ticks total nonlib name >>> > >>> > [GC]: >>> > ticks total nonlib name >>> > 0 0.0% >>> > >>> > [Call profile]: >>> > total call path >>> > 12 100.0% <no call path information> >>> > >>> > >>> > However, the v8.log file looks okay - it has 554 lines of data. >>> > >>> > What am I doing wrong? >>> >>> Could it be that the program is running for too short a time? Perhaps >>> if you try a longer-running js script you will get some ticks that are >>> hitting something interesting. >>> >>> Check that the paths to the executables and libraries that are listed >>> in the beginning of the v8.log file match something that the >>> tick-processor script can read and interpret with the nm command. >>> >>> -- >>> Erik Corry, Software Engineer >>> Google Denmark ApS. CVR nr. 28 86 69 84 >>> c/o Philip & Partners, 7 Vognmagergade, P.O. Box 2227, DK-1018 >>> Copenhagen K, Denmark. >>> >>> >>> >> >> >> > > > > --~--~---------~--~----~------------~-------~--~----~ v8-users mailing list [email protected] http://groups.google.com/group/v8-users -~----------~----~----~----~------~----~------~--~---
