the rw--- lines in the pmap output are the data segments. the r--- lines are code.
also have a look at the output of the "free" command. free is easier than top. -rob On Tue, Sep 7, 2010 at 1:15 PM, Michael Ganz <michael_g...@t-online.de> wrote: > > >> -----Ursprüngliche Nachricht----- >> Von: Michael Ganz [mailto:michael_g...@t-online.de] >> Gesendet: Dienstag, 7. September 2010 18:45 >> An: 'Rob Janes' >> Betreff: AW: preparing perl-interpreters >> >> >> >> > -----Ursprüngliche Nachricht----- >> > Von: Rob Janes [mailto:janes....@gmail.com] >> > Gesendet: Dienstag, 7. September 2010 17:16 >> > An: perl-xs@perl.org >> > Betreff: Re: preparing perl-interpreters >> > >> > it's possible perl is statically linked. ldd `type -p perl` will >> show >> > the libperl.so shared library if it's dynamically linked. >> >> It's linked shared, ldd shows an entry to libperl.so (Perl 5.12.1). >> >> > >> > valgrind will show actual memory usage. >> >> I have not yet valgrinded. OpenSuse 11.2 has an error here. But I try >> this on another box. >> >> > >> > pmap will show shared library segments. you can assume libperl.so >> > code segments are shared. i don't think the data segments are >> shared. >> >> pmap shows a lot of anon blocks with size of 2048K. >> 00002ad3ff039000 2048K 8K 4K 8K 0K rw-p [anon] >> 00002ad3ff239000 4K 0K 0K 0K 0K ---p >> [anon]00002ad404000000 131072K 130936K 11903K 130936K 0K rw-p >> [anon] >> ... >> >> And some really big ones: >> >> 00002ad410000000 655360K 654952K 59541K 654952K 0K rw-p [anon] >> 00002ad43c000000 168068K 167964K 15285K 167964K 0K rw-p [anon] >> >> > >> > ps is deceptive in that it does not report shared memory savings. >> >> I looked to the processes by top >> Now loading about 1000 perl-scripts and having 10 forked children of my >> perl-server. >> The Perl-Interpreter Instances are created in the perl-server >> >> top - 18:35:35 up 13 days, 6:17, 2 users, load average: 0.00, 0.00, >> 0.00 >> Tasks: 227 total, 1 running, 226 sleeping, 0 stopped, 0 zombie >> Cpu(s): 0.2%us, 0.1%sy, 0.0%ni, 99.6%id, 0.1%wa, 0.0%hi, 0.0%si, >> 0.0%st >> Mem: 8184844k total, 6689408k used, 1495436k free, 554284k >> buffers >> Swap: 16779852k total, 0k used, 16779852k free, 3988704k cached >> >> PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND >> 18790 mganz 20 0 1459m 1.0g 16m S 0 12.8 0:10.70 perlServer >> 19292 mganz 20 0 1459m 1.0g 3744 S 0 12.7 0:00.13 perlServer >> 19295 mganz 20 0 1459m 1.0g 3264 S 0 12.7 0:00.10 perlServer >> 19298 mganz 20 0 1459m 1.0g 3264 S 0 12.7 0:00.10 perlServer >> 19301 mganz 20 0 1459m 1.0g 3264 S 0 12.7 0:00.10 perlServer >> 19304 mganz 20 0 1459m 1.0g 3272 S 0 12.7 0:00.15 perlServer >> 19307 mganz 20 0 1459m 1.0g 3264 S 0 12.7 0:00.13 perlServer >> 19310 mganz 20 0 1459m 1.0g 3264 S 0 12.7 0:00.08 perlServer >> 19313 mganz 20 0 1459m 1.0g 3264 S 0 12.7 0:00.18 perlServer >> 19316 mganz 20 0 1459m 1.0g 3272 S 0 12.7 0:00.10 perlServer >> 19319 mganz 20 0 1459m 1.0g 3272 S 0 12.7 0:00.12 perlServer >> >> > >> > >> http://www.linuxquestions.org/linux/articles/Technical/Understanding_me >> > mory_usage_on_Linux >> > http://stackoverflow.com/questions/131303/linux-how-to-measure- >> actual- >> > memory-usage-of-an-application-or-process >> > >> > -rob >> > >> > On Tue, Sep 7, 2010 at 10:51 AM, Alexandre Jousset <m...@gtmp.org> >> > wrote: >> > > Hi, >> > > >> > > Le 07/09/2010 15:59, Michael Ganz a écrit : >> > >> >> > >> I saw that every interpreter-instance will allocate about 1MB of >> RAM >> > >> (Linux). >> > >> >> > >> So dealing with about 250 scripts, this allocates about 250 MB of >> > RAM. >> > > >> > > I may be wrong but if I understand correctly, it's 250MB of >> > *virtual* >> > > memory. Not real. libperl.so is loaded once and shared between >> > multiple >> > > instances. >> > > >> > > My 2c, >> > > -- >> > > -- \^/ -- >> > > -- -/ O \--------------------------------------- -- >> > > -- | |/ \| Alexandre (Midnite) Jousset | -- >> > > -- -|___|--------------------------------------- -- >> > > > >
