|
From: Harry M. <hj...@ta...> - 2005-05-11 20:04:14
|
The KCachegrind docs refer to an app called "op2callgrind" to tranlating oprofile information to data compatible with kcachegrind. I've not been able to find it via google and it does not appear to be in the oprofile source tree, nor the callgrind source tree nor the kcachegrind source tree. Did I miss something? Does anyone have any further info on it? <quote> System wide profiling is only permitted to the root user, as all actions on the system can be observed. Therefore, the following has to be done as root. First, configure the profiling process, using the GUI oprof_start or the command line tool opcontrol. Standard configuration should be timer mode (TBS, see introduction). To start the measurement, run opcontrol -s. Then run the application you are interested in, and afterwards, do a opcontrol -d. This will write out the measurement results into files under directory /var/lib/oprofile/samples/. To be able to visualize the data in KCachegrind, do in an empty directory: opreport -gdf | op2callgrind This will produce a lot of files, one for every program which was running on the system. Each one can be loaded into KCachegrind on its own. </quote> |
|
From: Harry M. <hj...@ta...> - 2005-05-11 20:25:35
|
Answering my own post -
it's called op2calltree in the converters subdir of kcachegrind
^^^^
a leftover from when callgrind was called calltree.
Sorry for the noise.
hjm
Harry Mangalam wrote: (before looking arefully..)
>
> The KCachegrind docs refer to an app called "op2callgrind" to tranlating
> oprofile information to data compatible with kcachegrind. I've not been
> able to find it via google and it does not appear to be in the oprofile
> source tree, nor the callgrind source tree nor the kcachegrind source
> tree.
> Did I miss something? Does anyone have any further info on it?
>
>
> <quote>
> System wide profiling is only permitted to the root user, as all actions
> on the system can be observed. Therefore, the following has to be done as
> root. First, configure the profiling process, using the GUI oprof_start or
> the command line tool opcontrol. Standard configuration should be timer
> mode (TBS, see introduction). To start the measurement, run opcontrol -s.
> Then run the application you are interested in, and afterwards, do a
> opcontrol -d. This will write out the measurement results into files under
> directory /var/lib/oprofile/samples/. To be able to visualize the data in
> KCachegrind, do in an empty directory:
>
>
> opreport -gdf | op2callgrind
>
> This will produce a lot of files, one for every program which was running
> on the system. Each one can be loaded into KCachegrind on its own.
> </quote>
>
>
>
> -------------------------------------------------------
> This SF.Net email is sponsored by Oracle Space Sweepstakes
> Want to be the first software developer in space?
> Enter now for the Oracle Space Sweepstakes!
> http://ads.osdn.com/?ad_id=7393&alloc_id=16281&op=click
|
|
From: Josef W. <Jos...@gm...> - 2005-05-12 14:40:21
|
On Wednesday 11 May 2005 22:23, Harry Mangalam wrote: > Answering my own post - > it's called op2calltree in the converters subdir of kcachegrind > ^^^^ > a leftover from when callgrind was called calltree. Hi, the "calltree" in "op2calltree" refers to the format, which is called by the tool which first generated the format, which was calltree. So there is not really a connection between this converter script and callgrind. I did not want to call it "cachegrind" format, as it contains a few extensions. As this format is created by cachegrind, calltree, callgrind, and now even cover, we could search for another name. Ideas? It is more or less simply a ASCII format to relate tuples of 64bit-integers to source positions. As it is used in the context of an executed program, it contains profile information. But perhaps it is best to call it CG-format (for cachegrind or callgrind), and thus call the converter "op2cg". Josef > > Sorry for the noise. > hjm > > > > Harry Mangalam wrote: (before looking arefully..) > > > The KCachegrind docs refer to an app called "op2callgrind" to tranlating > > oprofile information to data compatible with kcachegrind. I've not been > > able to find it via google and it does not appear to be in the oprofile > > source tree, nor the callgrind source tree nor the kcachegrind source > > tree. > > Did I miss something? Does anyone have any further info on it? > > > > > > <quote> > > System wide profiling is only permitted to the root user, as all actions > > on the system can be observed. Therefore, the following has to be done as > > root. First, configure the profiling process, using the GUI oprof_start > > or the command line tool opcontrol. Standard configuration should be > > timer mode (TBS, see introduction). To start the measurement, run > > opcontrol -s. Then run the application you are interested in, and > > afterwards, do a opcontrol -d. This will write out the measurement > > results into files under directory /var/lib/oprofile/samples/. To be able > > to visualize the data in KCachegrind, do in an empty directory: > > > > > > opreport -gdf | op2callgrind > > > > This will produce a lot of files, one for every program which was running > > on the system. Each one can be loaded into KCachegrind on its own. > > </quote> > > > > > > > > ------------------------------------------------------- > > This SF.Net email is sponsored by Oracle Space Sweepstakes > > Want to be the first software developer in space? > > Enter now for the Oracle Space Sweepstakes! > > http://ads.osdn.com/?ad_id=7393&alloc_id=16281&op=click > > ------------------------------------------------------- > This SF.Net email is sponsored by Oracle Space Sweepstakes > Want to be the first software developer in space? > Enter now for the Oracle Space Sweepstakes! > http://ads.osdn.com/?ad_id=7393&alloc_id=16281&op=click > _______________________________________________ > Valgrind-users mailing list > Val...@li... > https://lists.sourceforge.net/lists/listinfo/valgrind-users |
|
From: Nicholas N. <nj...@cs...> - 2005-05-12 14:46:54
|
On Thu, 12 May 2005, Josef Weidendorfer wrote: > But perhaps it is best to call it CG-format (for cachegrind or callgrind), and > thus call the converter "op2cg". I'd call it the "callgrind format". Cachegrind's cg_annotate script cannot read that format, so it seems useful to preserve the distinction between "cachegrind format" and "callgrind format". But I guess it's up to you, Josef. N |
|
From: Josef W. <Jos...@gm...> - 2005-05-12 15:35:32
|
On Thursday 12 May 2005 16:46, Nicholas Nethercote wrote: > On Thu, 12 May 2005, Josef Weidendorfer wrote: > > But perhaps it is best to call it CG-format (for cachegrind or > > callgrind), and thus call the converter "op2cg". > > I'd call it the "callgrind format". Cachegrind's cg_annotate script > cannot read that format, so it seems useful to preserve the distinction > between "cachegrind format" and "callgrind format". OK. Yes, makes sense. Have you ever tought about using Cachegrind's format for Massif for consensus points? I.e. event names: "Number of user bytes allocated", "Number of red zone bytes allocated", "Number of bytes freed", I know that the relation should be to call chains, not simple source positions. Callgrind can relate events to call chains, too, and uses a concatenation of the function names in the chain (seperated by an apostrophe) for the output. Hmm... as cg_annotate would have to know about this, the separation char should be part of the format. Josef > But I guess it's up > to you, Josef. > > N > > > ------------------------------------------------------- > This SF.Net email is sponsored by Oracle Space Sweepstakes > Want to be the first software developer in space? > Enter now for the Oracle Space Sweepstakes! > http://ads.osdn.com/?ad_id=7393&alloc_id=16281&op=click > _______________________________________________ > Valgrind-users mailing list > Val...@li... > https://lists.sourceforge.net/lists/listinfo/valgrind-users |
|
From: Nicholas N. <nj...@cs...> - 2005-05-13 13:53:11
|
On Thu, 12 May 2005, Josef Weidendorfer wrote: > Have you ever tought about using Cachegrind's format for Massif for consensus > points? I.e. event names: "Number of user bytes allocated", "Number of red > zone bytes allocated", "Number of bytes freed", I have honestly never thought of doing that. > I know that the relation should be to call chains, not simple source > positions. Yes. Hmm, it could be useful. But it's not on my TODO list :) N |