fche changed the topic of #systemtap to: http://sourceware.org/systemtap; email systemtap@sourceware.org if answers here not timely, conversations may be logged
<linus2>
fche: I had found a amazing phenomenon when using stap! The main.cpp can be compiled separately using "gcc -O3 -g main.cpp". The performance is A. But if compiled with other *.cpp using "g++ -O3 -g *.cpp", the performance is B. B is 76 times slower than A.
<linus2>
I had found a amazing phenomenon when using stap!!!! The main.cpp can be compiled separately using "g++ -O3 -g main.cpp". The performance is A. But if compiled with other *.cpp using "g++ -O3 -g *.cpp", the performance is B. B is 76 times slower than A.
<fche>
you mean some main.cpp that #includes all the other .cpp files?
<linus2>
no
<linus2>
There are many files of *.cpp
<linus2>
I compile it using g++ -03 -g *.cpp and generate a.out
<linus2>
Then I use stap to run this a.out.
<linus2>
But there are only the main.cpp important. And the main.cpp can compile using "g++ -O3 -g main.cpp"
<linus2>
but the former is slower than the latter
<linus2>
fche: I personally think that when running a program with stap, stap itself affects performance.
<linus2>
and the influence is very big.
<fche>
stap's influence depends on what you probe, how frequently the probes fire, and what the probes do
<fche>
linus2, so as a back-of-the-envelope estimate, a single uprobe hit on a modern workstation costs on the order of 0.1 - 1 microsecond
<linus2>
fche: So I need to re-evaluate stap's accuracy.
<fche>
it's not an accuracy question exactly, it's a "probe effect" question.
<linus2>
ok
hpt has quit [Ping timeout: 256 seconds]
_whitelogger has joined #systemtap
przemoc has quit [*.net *.split]
przemoc has joined #systemtap
amerey has joined #systemtap
sscox has quit [Ping timeout: 256 seconds]
tromey has joined #systemtap
derek0883 has joined #systemtap
wcohen has quit [Remote host closed the connection]
wcohen has joined #systemtap
derek0883 has quit [Remote host closed the connection]
derek0883 has joined #systemtap
derek0883 has quit [Read error: Connection reset by peer]
derek0883 has joined #systemtap
sscox has joined #systemtap
wcohen is now known as wcohen|lunch
mjw has joined #systemtap
irker296 has joined #systemtap
<irker296>
systemtap: smakarov systemtap.git:master * release-4.3-38-g601778117 / bpf-translate.cxx tapset-utrace.cxx tapsets.h: PR26234: stapbpf should warn about unsupported utrace_derived_probes
derek0883 has quit [Remote host closed the connection]
derek0883 has joined #systemtap
sapatel has joined #systemtap
orivej has quit [Ping timeout: 240 seconds]
<irker296>
systemtap: wcohen systemtap.git:wcohen/rhbz1854277 * release-4.3-38-ga7266dbde / dtrace.in: Make dtrace generated code work with LTO
derek0883 has quit [Remote host closed the connection]
derek0883 has joined #systemtap
<irker296>
systemtap: wcohen systemtap.git:master * release-4.3-39-g8212024da / dtrace.in: Make dtrace generated code work with LTO
mjw has quit [Quit: Leaving]
tromey has quit [Quit: ERC (IRC client for Emacs 28.0.50)]
derek0883 has quit [Remote host closed the connection]
derek0883 has joined #systemtap
amerey has quit [Quit: Leaving]
derek0883 has quit [Remote host closed the connection]
derek0883 has joined #systemtap
derek0883 has quit [Read error: No route to host]
derek0883 has joined #systemtap
sapatel has quit [Remote host closed the connection]