perfsuite-users — Mailing list for users of PerfSuite software

[PerfSuite-users] Subject:, Re: Compilation Error on perfsuite-0.6.2b2

[Rick K. <rk...@nc...>]

Thomas,

Your suspicion is probably correct.  This particular issue was not caught because we did not have the g77 compiler installed on the Ubuntu machine that is most similar to yours.  I am guessing there is probably some language dialect option that will cause the compiler to recognize that construct, but do not know what it is offhand.  We will look into it.

Any suggestions from list subscribers who may have a more recent Fortran compiler (GNU 4.2.x ideal) on their system are welcome!

Thanks again for reporting these problems - will update as more is known.

Rick




------------------------------------------------------------------------

Subject:
Re: Compilation Error on perfsuite-0.6.2b2
From:
"Thomas Tanaka" <tho...@gm...>
Date:
Wed, 26 Nov 2008 09:48:33 -0800

To:
per...@li...


I suspect it doesn't like integer*8 start declaration on line 57 on
perfsuite-0.6.2b2/src/libperfsuite/tests/psf_rtc-test.f

Thanks,

Thomas

On Wed, Nov 26, 2008 at 9:40 AM, Thomas Tanaka <tho...@gm...> wrote:

> > Hello again,
> >
> > I have tried to make the new perfsuite-0.6.2b2 and the following is the error:
> >
> > f77 -I../../../src/libperfsuite -g -O2 -o psf_error-test
> > psf_error-test.o
> > -L/home/tanaka/Desktop/ALL/perfsuite-0.6.2b2/src/libperfsuite
> > -lperfsuite
> > /usr/lib/gcc/i486-linux-gnu/4.2.3/../../../../lib/libf2c.so: undefined
> > reference to `MAIN__'
> > collect2: ld returned 1 exit status
> > f77 -I../../../src/libperfsuite  -g -O2 -c -o psf_rtc-test.o psf_rtc-test.f
> >   MAIN rtctest:
> > Warning on line 104: local variable sleepsecs never used
> >   dowork:
> > psf_rtc-test.f: In function 'MAIN__':
> > psf_rtc-test.f:74: error: expected '=', ',', ';', 'asm' or
> > '__attribute__' before 'stop'
> > psf_rtc-test.f:74: error: 'stop' undeclared (first use in this function)
> > psf_rtc-test.f:74: error: (Each undeclared identifier is reported only once
> > psf_rtc-test.f:74: error: for each function it appears in.)
> > psf_rtc-test.f:76: error: expected '=', ',', ';', 'asm' or
> > '__attribute__' before 'start'
> > psf_rtc-test.f:76: error: 'start' undeclared (first use in this function)
> > psf_rtc-test.f:80: error: expected '=', ',', ';', 'asm' or
> > '__attribute__' before 'elapsed'
> > psf_rtc-test.f:80: error: 'elapsed' undeclared (first use in this function)
> > psf_rtc-test.f:82: error: expected ')' before '*' token
> > /usr/bin/f77: aborting compilation
> > make[4]: *** [psf_rtc-test.o] Error 25
> > make[3]: *** [all-recursive] Error 1
> > make[2]: *** [all-recursive] Error 1
> > make[1]: *** [all-recursive] Error 1
> > make: *** [all] Error 2
> >
> > My platform is Linux 2.6.27-rc5-mm1 on ubuntu 8.04
> >
> > Any direction is much appreciated. Thanks in advance.
> >
> > Regards,
> >
> > Thomas
> >
>   

Re: [PerfSuite-users] Compilation error on perfsuite-0.6.2b1

[Rick K. <rk...@nc...>]

Thomas,

Thank you for raising this issue.  We became aware of this as well as a 
couple of other problems after the 0.6.2b1 release was posted.  They 
have since been resolved, and are undergoing final tests prior to a new 
0.6.2b2 release, which should solve the problem you experienced.  I 
expect the new release to be posted at SourceForge and NCSA later today 
or tomorrow.

Rick

p.s. your original email was bounced due to the perfsuite-users 
restricted to members only.  Your report is appended below.

Thomas Tanaka wrote:
> Hi All,
>
> I tried to make perfsuite-0.6.2b1 and was getting the following error:
>
> /bin/bash ../../../libtool --tag=CC   --mode=link gcc  -g -O2
> -L../../../src/libperfsuite -lperfsuite  -o ps_cpuusage-test
> ps_cpuusage-test.o
> gcc -g -O2 -o ps_cpuusage-test ps_cpuusage-test.o
> -L/home/tanaka/Development/perfsuite-0.6.2b1/src/libperfsuite
> -lperfsuite
> /home/tanaka/Development/perfsuite-0.6.2b1/src/libperfsuite/libperfsuite.a(libperfsuite_a-timers.o):
> In function `ps_rtc':
> /home/tanaka/Development/perfsuite-0.6.2b1/src/libperfsuite/timers.c:415:
> undefined reference to `rdtscll'
> collect2: ld returned 1 exit status
> make[4]: *** [ps_cpuusage-test] Error 1
> make[3]: *** [all-recursive] Error 1
> make[2]: *** [all-recursive] Error 1
> make[1]: *** [all-recursive] Error 1
> make: *** [all] Error 2
>
> I am currently using Linux Kernel 2.6.27-rc5-mm1 on Ubuntu 8.04
>
> Any help is appreciated. Thanks in advance
>
> Regards,
>
> Thomas
>
>   

[PerfSuite-users] PerfSuite at Supercomputing '08

[Rick K. <rk...@nc...>]

Subscribers to this list may be interested in the following events at the upcoming Supercomputing '08 conference to be held in Austin, TX:

1. There is an all-day tutorial on the NSF SDCI POINT Project (of which PerfSuite is a component) on Monday, November 17.  Information about this tutorial:

http://scyourway.nacse.org/conference/view/tut136

2. There is a Birds-of-a-Feather session (BOF) entitled "Tools for High Productivity Supercomputing" that will be held on Tuesday, November 18 from 12:15PM - 01:15PM, in Room 8C.  Information about this BOF:

http://scyourway.nacse.org/conference/view/bof127

3. Also on Tuesday, November 18, there will be demonstrations and discussions in the booth of Juelich Supercomputing Centre (booth #2621) of software from the POINT project as well as tools from the European VI-HPS project.  Information about these demos:

http://www.fz-juelich.de/jsc/neues/sc08/toolsTalks

Feel free to come to any of these events. Hope to see you in Austin!

Rick

[PerfSuite-users] PerfSuite 0.6.2b1 released

[Rick K. <rk...@nc...>]

PerfSuite 0.6.2 beta 1 is now available. 
 
PerfSuite is a collection of performance analysis software that can 
assist with a variety of techniques useful for application software 
optimization and benchmarking. The current release provides support for 
analysis with hardware performance counters on Linux. 
 
This release includes the following updates: 
 
- Updated to support recent Intel and AMD processors 
- Reimplementation of internals for more robust handling of 
processor cache hierarchy 
 
There are also additional bug fixes and enhancements in this 
release; please see the CHANGES file for a complete listing. 
 
URL: http://perfsuite.sourceforge.net/

Rick

Re: [PerfSuite-users] problem working with psprocess

[Rick K. <rk...@nc...>]

Dinesh -

Regarding the difficulty executing psprocess: you may need to rerun 
configure and change the option --with-tdom=$HOME/local/tdom to 
--with-tdom=$HOME/local/tdom/lib, as that option does not refer to the 
tDOM top-level installation, but rather the library subdirectory after 
installation.  Please try that and let me know if you still have failures.

On the second question: the -p option to psrun should support OpenMP if 
the underlying threading mechanism is POSIX threads. Intel's compilers, 
for example, use Pthreads and things generally work as you'd expect.

Rick

dinesh reddy wrote:
> I am using perfsuite-0.6.2a6,tcl8.6.a1 and tdom
> I am working on x86-64 linux machine
> I have installed perfsuite using the following script
>
> ./configure --prefix=$HOME/local/perfsuite \
> --with-tdom=$HOME/local/tdom \
> --with-tclsh=$HOME/local/tcl/
> bin/tclsh8.6 \
> --disable-binutils
>
>
> where $HOME is /home/y2008/cs6397/shm04/
> and tcl, tdom were installed in /home/y2008/cs6397/shm04/local
>
> psrun command is working fine and i got xml files also
> but when i run psprocess on those files i get the following message,
>
> /home/y2008/cs6397/shm04/local/perfsuite/bin/psprocess: line 61: 
> /home/y2008/cs6397/shm04/local/tcl/bin/tclsh8.6--with-tdom=/home/y2008/cs6397/shm04/local/tdom: 
> No such file or directory
> /home/y2008/cs6397/shm04/local/perfsuite/bin/psprocess: line 61: exec: 
> /home/y2008/cs6397/shm04/local/tcl/bin/tclsh8.6--with-tdom=/home/y2008/cs6397/shm04/local/tdom: 
> cannot execute: No such file or directory
>
> and does -p option works for openMP programs also
> ------------------------------------------------------------------------
>
> -------------------------------------------------------------------------
> This SF.Net email is sponsored by the Moblin Your Move Developer's challenge
> Build the coolest Linux based applications with Moblin SDK & win great prizes
> Grand prize is a trip for two to an Open Source event anywhere in the world
> http://moblin-contest.org/redirect.php?banner_id=100&url=/
> ------------------------------------------------------------------------
>
> _______________________________________________
> PerfSuite-users mailing list
> Per...@li...
> https://lists.sourceforge.net/lists/listinfo/perfsuite-users
>   

[PerfSuite-users] problem working with psprocess

[dinesh r. <din...@gm...>]

I am using perfsuite-0.6.2a6,tcl8.6.a1 and tdom
I am working on x86-64 linux machine
I have installed perfsuite using the following script

./configure --prefix=$HOME/local/perfsuite \
--with-tdom=$HOME/local/tdom \
--with-tclsh=$HOME/local/tcl/bin/tclsh8.6 \
--disable-binutils


where $HOME is /home/y2008/cs6397/shm04/
and tcl, tdom were installed in /home/y2008/cs6397/shm04/local

psrun command is working fine and i got xml files also
but when i run psprocess on those files i get the following message,

/home/y2008/cs6397/shm04/local/perfsuite/bin/psprocess: line 61:
/home/y2008/cs6397/shm04/local/tcl/bin/tclsh8.6--with-tdom=/home/y2008/cs6397/shm04/local/tdom:
No such file or directory
/home/y2008/cs6397/shm04/local/perfsuite/bin/psprocess: line 61: exec:
/home/y2008/cs6397/shm04/local/tcl/bin/tclsh8.6--with-tdom=/home/y2008/cs6397/shm04/local/tdom:
cannot execute: No such file or directory

and does -p option works for openMP programs also

Re: [PerfSuite-users] Problems running examples

[Rick K. <rk...@nc...>]

Abdullah -

Please try to use the SourceForge mailing list for questions/comments 
about PerfSuite.  That allows traffic to be archived that may benefit 
other people who have similar questions by browsing earlier postings.  
It also helps me keep my NCSA (work) email distinct, which in turn keeps 
my employer happy :)

As far as your question goes: much is dependent on the particular MPI 
implementation and how it manages processes/tasks.  For example, MPICH 
using the p4 device won't work with psrun at all (this was discussed in 
a perfsuite-users posting a long time ago).   You may have to experiment 
with your own favorite MPI.   Charm++ is another case that can work in 
some cases, others not.

The driver for the processor's performance monitoring unit (perfmon or 
perfctr for Linux) takes care of virtualizing the hardware counter data 
on a per-thread basis, so if your application spawns separate processes 
that use multiple cores, they should each generate an output XML 
document.  For multithreaded applications that use POSIX threads (either 
directly or through a Pthread-based interface like many OpenMP 
implementations), the option "-p" that psrun accepts will trap thread 
creation and generate per-thread performance data at thread exit.

If you are instead looking for a system-wide monitoring tool that would 
track all cores over a given period of time, that is not what PerfSuite 
was intended to do, but there are other tools available that fill that 
space.

Rick

Abdullah Kayi wrote:
> Hi Rick,
>
> I am pretty much getting the reasonable outputs from the tests that I 
> tried. So now, I am trying to use "psrun" with mpirun I couldn't find 
> an example for that other then cpi-mpi which works with libpshwpc. So 
> I  am not sure what is needed for psrun to work for mpi or charm++ 
> directly. Also, how can I get theoutput from all the cores? Do you 
> have any ideas on these? Thanks so much for your help.
>
> Regards,
>
> Abdullah
>
> On Wed, Jun 25, 2008 at 5:39 PM, Abdullah Kayi <ap...@gm... 
> <mailto:ap...@gm...>> wrote:
>
>     Ooops my bad, I shouldn't have used "psrun", now I am getting the
>     correct results. I will let you know about the ChaNGa results.
>
>     On Wed, Jun 25, 2008 at 4:14 PM, Abdullah Kayi <ap...@gm...
>     <mailto:ap...@gm...>> wrote:
>
>         Hey Rick,
>
>         I am running the "hl" example like this;
>
>
>         psrun -c papi3_mflops.xml ./hl
>
>         I am finally getting the xml output  fine but when I look at
>         the results.
>
>         Index
>         Description                                                             
>         Counter Value
>         ============================================================================================
>             1 Floating point
>         operations........................................                   
>         3
>             2 Total
>         cycles.....................................................              
>         629519
>
>         Event Index
>         ============================================================================================
>             1: PAPI_FP_OPS         2: PAPI_TOT_CYC
>
>         There should be something wrong how come there were only 3 FP
>         operations in that output, I am using the "hl.c" as it is. Do
>         you have any ideas on this?
>
>         Regards,
>
>         Abdullah
>
>
>

Re: [PerfSuite-users] Problems running examples

[Rick K. <rk...@nc...>]

Abdullah -

It would appear that all of the errors you are encountering are related 
to use of PAPI calls within PerfSuite, so I would like to step back and 
make sure that the underlying software is functioning properly.  To help 
do that, I have some basic questions:

Can you clarify what processor you are using?  The config.log output you 
sent indicates it is an x86_64 machine, but something more specific 
would be helpful.   "cat /proc/cpuinfo" or the output of the psinv 
command may help.

Is this a perfctr or perfmon-patched system?

I'm also guessing you are using PAPI version 3.5.0, based on the 
pathname you provided via --with-papi on the configure command line.  
Unfortunately, I'm not sure if that is the official PAPI 3.5.0 release 
or possibly a version updated later via CVS.

Did you run the PAPI test suite when you built the library ("make 
fulltest" within the PAPI src subdirectory)?  If so, did all tests pass 
as expected?

Did you run the PerfSuite test suite as well ("make check")?  How about 
the outcome of those tests?  These are helpful tests to run as they help 
ensure that PAPI is functioning correctly outside of PerfSuite, 
particularly the test cases in the "libpshwpc" group.

If any of the above failed, then it would be helpful to see the relevant 
output before considering the examples as well as anything more 
substantial (like Charm++ runs).

Rick

Abdullah Kayi wrote:
> And on top of these, for example if I try to run "hl" with "psrun -c 
> papi3_mflops.xml ./hl"
>
> I am getting the following error;
>
>
> hwpc.c:306 : Initialized global data
> hwpc.c:315 : Signal handling off.
> hwpc.c:2288 : Configuration file set to papi3_mflops.xml
> hwpc.c:2333 : Attempting to parse "papi3_mflops.xml"
> xmlparse.c:301 : XML parser: parsing expected XML document element 
> ps_hwpc_eventlist (class = "PAPI")
> xmlparse.c:237 : XML parser: found preset: "PAPI_FP_OPS"
> xmlparse.c:237 : XML parser: found preset: "PAPI_TOT_CYC"
> xmlparse.c:490 : XML parser: found 2 events:
>         PAPI_FP_OPS
>         PAPI_TOT_CYC
> hwpc.c:323 : Configuration file read
> hwpc.c:325 : Read 2 events
> hwpc.c:2158 : Initializing PAPI class...
> hwpc.c:354 : Performance software supports 32 events max.
> hwpc.c:447 : Calling package init routine
> hwpc-papi.c:319 : PAPI/PerfSuite version check failed (require at 
> least 3, got -3)
> libpsrun fatal error: error reported by performance software layer
>
> Thanks in advance,
>
> Regards,
>
> Abdullah
>
> On Tue, Jun 24, 2008 at 4:33 PM, Abdullah Kayi <ap...@gm... 
> <mailto:ap...@gm...>> wrote:
>
>     Hi Rick,
>
>     You can ignore number #1 as I found the problem with my
>     configuration. Any help on the remaining issues are appreciated.
>
>     Regards,
>
>     Abdullah
>
>
>     On Tue, Jun 24, 2008 at 3:07 PM, Abdullah Kayi <ap...@gm...
>     <mailto:ap...@gm...>> wrote:
>
>         Hi Rick,
>
>         I made a clean install. Here is some info from the config.log
>         and I solved some of the problems that I had with the examples.
>
>         ./configure --prefix=/wrkdir/kayi/PERFSUITE_DIR
>         --with-papi=/u2/wes/PET_HOME/pkgs/papi-3.5.0
>         --with-tdom=/wrkdir/kayi/TDOM_DIR PTHREAD_LIBS=-lpthread
>         --enable-mpi
>         MPICPPFLAGS=-I/usr/local/pkg/voltairempi/voltairempi-S-1/mpi.pathcc.rsh/include
>         --enable-debug
>
>         ## --------- ##
>         ## Platform. ##
>         ## --------- ##
>
>         uname -m = x86_64
>         uname -r = 2.6.5-7.276-smp
>         uname -s = Linux
>
>         However, I still have some questions and issues.
>
>         1- First of all, the "psprocess" wasn't installed although I
>         have the tcl and tdom in my system. So am I missing any
>         requirements for that to install?  how can i parse the xml
>         output otherwise?
>
>         2- If I run the "cpi" with "psrun -c papi_profile_l2tcm.xml
>         ./cpi" I am getting the following error with PS_DEBUG=10
>
>         hwpc-papi.c:750 : Initializing profiling in PAPI layer
>         hwpc-papi.c:152 : Entering setup_counters_internal
>         hwpc-papi.c:193 : Eventset address: 0x504070
>         hwpc-papi.c:209 : Created an eventset (0).
>         hwpc-papi.c:228 : Domain of eventset: unknown (0)
>         hwpc-papi.c:270 : Added 1 events to eventset 0
>         hwpc-papi.c:777 : Profiling using PAPI event code 80000007
>         (PAPI_L2_TCM)
>         hwpc-papi.c:789 : PAPI_sprofil_t structure allocated at 0x504bc0
>         hwpc-papi.c:810 : Calling PAPI_sprofil with arguments
>         sprofils  = 0x504bc0
>         nmaps     = 11
>         *eventset = 0
>         eventcode = 80000007
>         threshold = 100000
>         flags     = 0
>         hwpc-papi.c:818 : PAPI_sprofil error: PAPI_EINVAL
>         hwpc.c:602 : Error return from init_profiling
>
>         libpsrun fatal error: error reported by performance software layer
>
>         If I run the mpirun version I am also getting the same error.
>         So how can I enable the profiling runs?
>
>         3- if I use psrun with mpirun (or charmrun) on a Charm++ code
>         ChaNGa I am getting the following error:
>
>          
>         PAPI/PerfSuite version check failed (require at least 3, got -3)
>
>         libpsrun fatal error: error reported by performance software layer
>         [psrun: Parent process forked, PID=1329]
>         [psrun: Child process forked, PID=2976]
>         hwpc-papi.c:718 : PAPI_shutdown called
>         hwpc.c:1046 : ps_hwpc_shutdown: returned from destroy counters
>         hwpc.c:1733 : No PID file scheduled for deletion.
>         fork(): _psrun_hwpc_shutdown returned 0
>         [psrun: Parent process forked, PID=2976]
>         [psrun: Parent process forked, PID=2976]
>         [psrun: Child process forked, PID=2977]
>         [psrun: Child process forked, PID=2978]
>
>
>         Do you have any idea what might be wrong in these different
>         issues? Thanks so much for your help and consideration.
>
>         Regards,
>
>         Abdullah
>
>
>
> ------------------------------------------------------------------------
>
> -------------------------------------------------------------------------
> Check out the new SourceForge.net Marketplace.
> It's the best place to buy or sell services for
> just about anything Open Source.
> http://sourceforge.net/services/buy/index.php
> ------------------------------------------------------------------------
>
> _______________________________________________
> PerfSuite-users mailing list
> Per...@li...
> https://lists.sourceforge.net/lists/listinfo/perfsuite-users
>   

Re: [PerfSuite-users] Problems running examples

[Abdullah K. <ap...@gm...>]

And on top of these, for example if I try to run "hl" with "psrun -c
papi3_mflops.xml ./hl"

I am getting the following error;


hwpc.c:306 : Initialized global data
hwpc.c:315 : Signal handling off.
hwpc.c:2288 : Configuration file set to papi3_mflops.xml
hwpc.c:2333 : Attempting to parse "papi3_mflops.xml"
xmlparse.c:301 : XML parser: parsing expected XML document element
ps_hwpc_eventlist (class = "PAPI")
xmlparse.c:237 : XML parser: found preset: "PAPI_FP_OPS"
xmlparse.c:237 : XML parser: found preset: "PAPI_TOT_CYC"
xmlparse.c:490 : XML parser: found 2 events:
        PAPI_FP_OPS
        PAPI_TOT_CYC
hwpc.c:323 : Configuration file read
hwpc.c:325 : Read 2 events
hwpc.c:2158 : Initializing PAPI class...
hwpc.c:354 : Performance software supports 32 events max.
hwpc.c:447 : Calling package init routine
hwpc-papi.c:319 : PAPI/PerfSuite version check failed (require at least 3,
got -3)
libpsrun fatal error: error reported by performance software layer

Thanks in advance,

Regards,

Abdullah

On Tue, Jun 24, 2008 at 4:33 PM, Abdullah Kayi <ap...@gm...> wrote:

> Hi Rick,
>
> You can ignore number #1 as I found the problem with my configuration. Any
> help on the remaining issues are appreciated.
>
> Regards,
>
> Abdullah
>
>
> On Tue, Jun 24, 2008 at 3:07 PM, Abdullah Kayi <ap...@gm...> wrote:
>
>> Hi Rick,
>>
>> I made a clean install. Here is some info from the config.log and I solved
>> some of the problems that I had with the examples.
>>
>> ./configure --prefix=/wrkdir/kayi/PERFSUITE_DIR
>> --with-papi=/u2/wes/PET_HOME/pkgs/papi-3.5.0
>> --with-tdom=/wrkdir/kayi/TDOM_DIR PTHREAD_LIBS=-lpthread --enable-mpi
>> MPICPPFLAGS=-I/usr/local/pkg/voltairempi/voltairempi-S-1/mpi.pathcc.rsh/include
>> --enable-debug
>>
>> ## --------- ##
>> ## Platform. ##
>> ## --------- ##
>>
>> uname -m = x86_64
>> uname -r = 2.6.5-7.276-smp
>> uname -s = Linux
>>
>> However, I still have some questions and issues.
>>
>> 1- First of all, the "psprocess" wasn't installed although I have the tcl
>> and tdom in my system. So am I missing any requirements for that to
>> install?  how can i parse the xml output otherwise?
>>
>> 2- If I run the "cpi" with "psrun -c papi_profile_l2tcm.xml ./cpi" I am
>> getting the following error with PS_DEBUG=10
>>
>> hwpc-papi.c:750 : Initializing profiling in PAPI layer
>> hwpc-papi.c:152 : Entering setup_counters_internal
>> hwpc-papi.c:193 : Eventset address: 0x504070
>> hwpc-papi.c:209 : Created an eventset (0).
>> hwpc-papi.c:228 : Domain of eventset: unknown (0)
>> hwpc-papi.c:270 : Added 1 events to eventset 0
>> hwpc-papi.c:777 : Profiling using PAPI event code 80000007 (PAPI_L2_TCM)
>> hwpc-papi.c:789 : PAPI_sprofil_t structure allocated at 0x504bc0
>> hwpc-papi.c:810 : Calling PAPI_sprofil with arguments
>> sprofils  = 0x504bc0
>> nmaps     = 11
>> *eventset = 0
>> eventcode = 80000007
>> threshold = 100000
>> flags     = 0
>> hwpc-papi.c:818 : PAPI_sprofil error: PAPI_EINVAL
>> hwpc.c:602 : Error return from init_profiling
>> libpsrun fatal error: error reported by performance software layer
>>
>> If I run the mpirun version I am also getting the same error. So how can I
>> enable the profiling runs?
>>
>> 3- if I use psrun with mpirun (or charmrun) on a Charm++ code ChaNGa I am
>> getting the following error:
>>
>>
>> PAPI/PerfSuite version check failed (require at least 3, got -3)
>> libpsrun fatal error: error reported by performance software layer
>>  [psrun: Parent process forked, PID=1329]
>> [psrun: Child process forked, PID=2976]
>> hwpc-papi.c:718 : PAPI_shutdown called
>> hwpc.c:1046 : ps_hwpc_shutdown: returned from destroy counters
>> hwpc.c:1733 : No PID file scheduled for deletion.
>> fork(): _psrun_hwpc_shutdown returned 0
>> [psrun: Parent process forked, PID=2976]
>> [psrun: Parent process forked, PID=2976]
>> [psrun: Child process forked, PID=2977]
>> [psrun: Child process forked, PID=2978]
>>
>>
>> Do you have any idea what might be wrong in these different issues? Thanks
>> so much for your help and consideration.
>>
>> Regards,
>>
>> Abdullah
>>
>
>

Re: [PerfSuite-users] Problems running examples

[Abdullah K. <ap...@gm...>]

Hi Rick,

You can ignore number #1 as I found the problem with my configuration. Any
help on the remaining issues are appreciated.

Regards,

Abdullah

On Tue, Jun 24, 2008 at 3:07 PM, Abdullah Kayi <ap...@gm...> wrote:

> Hi Rick,
>
> I made a clean install. Here is some info from the config.log and I solved
> some of the problems that I had with the examples.
>
> ./configure --prefix=/wrkdir/kayi/PERFSUITE_DIR
> --with-papi=/u2/wes/PET_HOME/pkgs/papi-3.5.0
> --with-tdom=/wrkdir/kayi/TDOM_DIR PTHREAD_LIBS=-lpthread --enable-mpi
> MPICPPFLAGS=-I/usr/local/pkg/voltairempi/voltairempi-S-1/mpi.pathcc.rsh/include
> --enable-debug
>
> ## --------- ##
> ## Platform. ##
> ## --------- ##
>
> uname -m = x86_64
> uname -r = 2.6.5-7.276-smp
> uname -s = Linux
>
> However, I still have some questions and issues.
>
> 1- First of all, the "psprocess" wasn't installed although I have the tcl
> and tdom in my system. So am I missing any requirements for that to
> install?  how can i parse the xml output otherwise?
>
> 2- If I run the "cpi" with "psrun -c papi_profile_l2tcm.xml ./cpi" I am
> getting the following error with PS_DEBUG=10
>
> hwpc-papi.c:750 : Initializing profiling in PAPI layer
> hwpc-papi.c:152 : Entering setup_counters_internal
> hwpc-papi.c:193 : Eventset address: 0x504070
> hwpc-papi.c:209 : Created an eventset (0).
> hwpc-papi.c:228 : Domain of eventset: unknown (0)
> hwpc-papi.c:270 : Added 1 events to eventset 0
> hwpc-papi.c:777 : Profiling using PAPI event code 80000007 (PAPI_L2_TCM)
> hwpc-papi.c:789 : PAPI_sprofil_t structure allocated at 0x504bc0
> hwpc-papi.c:810 : Calling PAPI_sprofil with arguments
> sprofils  = 0x504bc0
> nmaps     = 11
> *eventset = 0
> eventcode = 80000007
> threshold = 100000
> flags     = 0
> hwpc-papi.c:818 : PAPI_sprofil error: PAPI_EINVAL
> hwpc.c:602 : Error return from init_profiling
> libpsrun fatal error: error reported by performance software layer
>
> If I run the mpirun version I am also getting the same error. So how can I
> enable the profiling runs?
>
> 3- if I use psrun with mpirun (or charmrun) on a Charm++ code ChaNGa I am
> getting the following error:
>
>
> PAPI/PerfSuite version check failed (require at least 3, got -3)
> libpsrun fatal error: error reported by performance software layer
> [psrun: Parent process forked, PID=1329]
> [psrun: Child process forked, PID=2976]
> hwpc-papi.c:718 : PAPI_shutdown called
> hwpc.c:1046 : ps_hwpc_shutdown: returned from destroy counters
> hwpc.c:1733 : No PID file scheduled for deletion.
> fork(): _psrun_hwpc_shutdown returned 0
> [psrun: Parent process forked, PID=2976]
> [psrun: Parent process forked, PID=2976]
> [psrun: Child process forked, PID=2977]
> [psrun: Child process forked, PID=2978]
>
>
> Do you have any idea what might be wrong in these different issues? Thanks
> so much for your help and consideration.
>
> Regards,
>
> Abdullah
>

Re: [PerfSuite-users] Problems running examples

[Abdullah K. <ap...@gm...>]

Hi Rick,

I made a clean install. Here is some info from the config.log and I solved
some of the problems that I had with the examples.

./configure --prefix=/wrkdir/kayi/PERFSUITE_DIR
--with-papi=/u2/wes/PET_HOME/pkgs/papi-3.5.0
--with-tdom=/wrkdir/kayi/TDOM_DIR PTHREAD_LIBS=-lpthread --enable-mpi
MPICPPFLAGS=-I/usr/local/pkg/voltairempi/voltairempi-S-1/mpi.pathcc.rsh/include
--enable-debug

## --------- ##
## Platform. ##
## --------- ##

uname -m = x86_64
uname -r = 2.6.5-7.276-smp
uname -s = Linux

However, I still have some questions and issues.

1- First of all, the "psprocess" wasn't installed although I have the tcl
and tdom in my system. So am I missing any requirements for that to
install?  how can i parse the xml output otherwise?

2- If I run the "cpi" with "psrun -c papi_profile_l2tcm.xml ./cpi" I am
getting the following error with PS_DEBUG=10

hwpc-papi.c:750 : Initializing profiling in PAPI layer
hwpc-papi.c:152 : Entering setup_counters_internal
hwpc-papi.c:193 : Eventset address: 0x504070
hwpc-papi.c:209 : Created an eventset (0).
hwpc-papi.c:228 : Domain of eventset: unknown (0)
hwpc-papi.c:270 : Added 1 events to eventset 0
hwpc-papi.c:777 : Profiling using PAPI event code 80000007 (PAPI_L2_TCM)
hwpc-papi.c:789 : PAPI_sprofil_t structure allocated at 0x504bc0
hwpc-papi.c:810 : Calling PAPI_sprofil with arguments
sprofils  = 0x504bc0
nmaps     = 11
*eventset = 0
eventcode = 80000007
threshold = 100000
flags     = 0
hwpc-papi.c:818 : PAPI_sprofil error: PAPI_EINVAL
hwpc.c:602 : Error return from init_profiling
libpsrun fatal error: error reported by performance software layer

If I run the mpirun version I am also getting the same error. So how can I
enable the profiling runs?

3- if I use psrun with mpirun (or charmrun) on a Charm++ code ChaNGa I am
getting the following error:


PAPI/PerfSuite version check failed (require at least 3, got -3)
libpsrun fatal error: error reported by performance software layer
[psrun: Parent process forked, PID=1329]
[psrun: Child process forked, PID=2976]
hwpc-papi.c:718 : PAPI_shutdown called
hwpc.c:1046 : ps_hwpc_shutdown: returned from destroy counters
hwpc.c:1733 : No PID file scheduled for deletion.
fork(): _psrun_hwpc_shutdown returned 0
[psrun: Parent process forked, PID=2976]
[psrun: Parent process forked, PID=2976]
[psrun: Child process forked, PID=2977]
[psrun: Child process forked, PID=2978]


Do you have any idea what might be wrong in these different issues? Thanks
so much for your help and consideration.

Regards,

Abdullah

Re: [PerfSuite-users] Problems running examples

[Rick K. <rk...@nc...>]

Abdullah,

This is somewhat of a "catch-all"/generic error that can be a little 
hard to diagnose without additional information, such as what type of 
hardware you are running on, which version of PAPI you are using, which 
kernel support you have (perfctr or perfmon).

There are occasional posts in this mailing list from others that may be 
useful to have a look at, see for example:

http://sourceforge.net/mailarchive/forum.php?thread_name=20061117154330.GA25518%40ichec.ie&forum_name=perfsuite-users

For me, it is typically best do reconfigure/recompile PerfSuite with 
debugging enabled ("--enable-debug"), try to run an example with the 
environment variable PS_DEBUG set to the value 3 or higher, and examine 
the debugging output to learn where things are failing (or pass that on 
to me, as it's easier for me to make sense of it).

Rick

p.s. it has been a very long time since an update to PerfSuite has been 
released, but one currently being tested should hopefully be available 
within the next couple of weeks: 0.6.2a7

>
>
> Subject:
> Problems running examples
> From:
> "Abdullah Kayi" <ap...@gm...>
> Date:
> Mon, 23 Jun 2008 18:41:35 -0800
> To:
> per...@li...
>
> To:
> per...@li...
>
>
> Hi all,
>
> I am using PerfSuite 0.6.2a6. I compiled and installed succesfully but 
> I cannot run any of the examples. I am always getting the following error:
> libpsrun fatal error: error reported by performance software layer
>
> Any ideas on this?
>
> Regards,
>
> Abdullah

Re: [PerfSuite-users] Does Perfsuite support Perfmon2 on Core 2 Quad cpu?

[Rick K. <rk...@nc...>]

Chiaheng,

This is a long overdue update to PerfSuite that you mention.  The last 
released version does not have the updates necessary for the Intel 
Core/Core 2 systems, but this will soon be addressed.  We are currently 
running and testing on NCSA's Clovertown cluster a pre-release version 
of the next release which will have support for these processors as well 
as AMD family k10.  The release will support PAPI running on either 
perfctr- or perfmon-2 patched systems.

So - the answer to your question is that PerfSuite 0.6.2a6 does not 
support these processors, but 0.6.2a7 will.  Apologies to all for the 
lengthy delay between releases, but a new one is at hand (as is a new 
release of PAPI, version 3.6).

Rick

per...@li... wrote:
> The attached message has been automatically discarded.
>
> ------------------------------------------------------------------------
>
> Subject:
> Does Perfsuite support Perfmon2 on Core 2 Quad cpu?
> From:
> chiaheng <d94...@cs...>
> Date:
> Mon, 28 Apr 2008 21:57:12 +0800
> To:
> per...@li...
>
> To:
> per...@li...
>
>
> Dear All,
>
>    This is chiaheng. I would like to your help to know that if 
> Perfsuite supports PAPI running with Perfmon2 on Core 2 Quad machine???
>
> Thanks in advance,
> Chiaheng.
>

[PerfSuite-users] POINT project announcement

[Rick K. <rk...@nc...>]

Greetings -

The following is an announcement appearing in this week's issue of 
HPCwire, introducing the NSF-funded POINT project, of which PerfSuite is 
a participant:

http://www.hpcwire.com/hpc/2232188.html

Rick

[PerfSuite-users] PerfSuite developer position search extended

[Rick K. <rk...@nc...>]

Please note: we have extended the date range for which applications will 
be accepted for the new full-time position at NCSA/Illinois to assist in 
PerfSuite development and support for performance analysis software on 
current and next generation HPC systems.

Please feel free to pass along this notice to any colleagues who may 
have interest and/or experience in the area of high-performance 
computing.  I encourage interested people to browse both the NCSA and 
PerfSuite websites to learn about recent developments at the center, 
including work towards the upcoming Blue Waters petascale computing 
system for open scientific research.

The URL given below is still valid for further information.

Rick

Rick Kufrin wrote:
> Greetings -
>
> Thanks to new funding support from the National Science Foundation, 
> there is an opportunity to join in development of PerfSuite as described 
> in the following employment posting from NCSA:
>
> http://www.ncsa.uiuc.edu/AboutUs/Employment/11659.html
>
> This is an exciting development that should help accelerate the 
> evolution of PerfSuite greatly.  Please feel free to share this 
> information with anyone who may be interested in helping develop this 
> open source project (and yes, get paid for it!).  Note: job posting 
> closes on March 14.
>
> Rick
>
> -------------------------------------------------------------------------
> This SF.net email is sponsored by: Microsoft
> Defy all challenges. Microsoft(R) Visual Studio 2008.
> http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/
> _______________________________________________
> PerfSuite-users mailing list
> Per...@li...
> https://lists.sourceforge.net/lists/listinfo/perfsuite-users
>
>   

Re: [PerfSuite-users] Floating Point Instructions

[Rick K. <rk...@nc...>]

Abel -

> What's the difference between a floating point instruction and a 
> floating point operation? I have papi installed with perfsuite on top 
> of it. When I run a program I get a large number for floating point 
> operations but zero for floating point instructions. First if all I 
> don't know the difference and second of all why am getting zero for 
> floating point instructions? Here is the output to psrun.
>
>

I think the following PAPI FAQ may help address this question:

http://icl.cs.utk.edu/papi/faq/index.html#213

Rick

[PerfSuite-users] Floating Point Instructions

[Abel L. <abe...@gm...>]

What's the difference between a floating point instruction and a  
floating point operation? I have papi installed with perfsuite on top  
of it. When I run a program I get a large number for floating point  
operations but zero for floating point instructions. First if all I  
don't know the difference and second of all why am getting zero for  
floating point instructions? Here is the output to psrun.

<hwpcevent type="preset" name="PAPI_BR_INS" derived="no">996217933</ 
hwpcevent>
                 <hwpcevent type="preset" name="PAPI_BR_MSP"  
derived="no">2639</hwpcevent>
                 <hwpcevent type="preset" name="PAPI_BR_TKN"  
derived="no">995330538</hwpcevent>
                 <hwpcevent type="preset" name="PAPI_FP_INS"  
derived="yes">0</hwpcevent>
                 <hwpcevent type="preset" name="PAPI_FP_OPS"  
derived="yes">2983143200</hwpcevent>
                 <hwpcevent type="preset" name="PAPI_L1_ICA"  
derived="no">26839719801</hwpcevent>
                 <hwpcevent type="preset" name="PAPI_L1_ICM"  
derived="no">35060</hwpcevent>
                 <hwpcevent type="preset" name="PAPI_L2_TCH"  
derived="no">249722</hwpcevent>
                 <hwpcevent type="preset" name="PAPI_L2_TCM"  
derived="no">59</hwpcevent>
                 <hwpcevent type="preset" name="PAPI_RES_STL"  
derived="no">0</hwpcevent>
                 <hwpcevent type="preset" name="PAPI_TLB_DM"  
derived="no">21720</hwpcevent>
                 <hwpcevent type="preset" name="PAPI_TLB_IM"  
derived="no">7835</hwpcevent>
                 <hwpcevent type="preset" name="PAPI_TLB_TL"  
derived="no">30942</hwpcevent>
                 <hwpcevent type="preset" name="PAPI_TOT_CYC"  
derived="no">82809635010</hwpcevent>
                 <hwpcevent type="preset" name="PAPI_TOT_INS"  
derived="no">12960989996</hwpcevent>
                 <hwpcevent type="preset" name="PAPI_VEC_INS"  
derived="yes">4982674130</hwpcevent>


Here is  psinv -p
System Information -
Node Name:              bhutan
OS Name:                Linux
OS Release:             2.6.13-perfctr-15.18-default
OS Build/Version:       #1 Wed Feb 6 22:58:43 EST 2008
OS Machine:             x86_64
Processors:             1
Total Memory (MB):      1000.40
System Page Size (KB):  4.00

Processor Information -
Vendor:                 Intel
Processor family:       Pentium 4
Brand:                  Intel(R) Pentium(R) 4 CPU 3.00GHz
Model (Type):           (unknown)
Revision:               3
Clock Speed:            2992.61 MHz

Cache and TLB Information -
Cache levels:           2
Caches/TLBs:            5

Cache Details -
Level 1:
         Type:           Data
         Size:           16 KB
         Line size:      64 bytes
         Associativity:  8-way set associative

         Type:           Instruction Trace
         Size:           12K uOps
         Associativity:  8-way set associative

Level 2:
         Type:           Unified
         Size:           2.00 MB
         Line size:      64 bytes
         Associativity:  8-way set associative

TLB Details -
Level 1:
         Type:           Instruction
         Entries:        64
         Pagesize (KB):  4 2048 4096
         Associativity:  Fully associative

         Type:           Data
         Entries:        64
         Pagesize (KB):  4 4096
         Associativity:  Fully associative

PAPI Version Information -
Major version:          3
Minor version:          5
Revision:               0

PAPI Standard Event Information -
Standard events:        23
Non-derived events:     17
Derived events:         6

PAPI Standard Event Details -
Non-derived:
         PAPI_BR_INS:    Branch instructions
         PAPI_BR_MSP:    Conditional branch instructions mispredicted
         PAPI_BR_NTK:    Conditional branch instructions not taken
         PAPI_BR_PRC:    Conditional branch instructions correctly  
predicted
         PAPI_BR_TKN:    Conditional branch instructions taken
         PAPI_L1_ICA:    Level 1 instruction cache accesses
         PAPI_L1_ICM:    Level 1 instruction cache misses
         PAPI_L2_TCA:    Level 2 total cache accesses
         PAPI_L2_TCH:    Level 2 total cache hits
         PAPI_L2_TCM:    Level 2 cache misses
         PAPI_RES_STL:   Cycles stalled on any resource
         PAPI_TLB_DM:    Data translation lookaside buffer misses
         PAPI_TLB_IM:    Instruction translation lookaside buffer misses
         PAPI_TLB_TL:    Total translation lookaside buffer misses
         PAPI_TOT_CYC:   Total cycles
         PAPI_TOT_IIS:   Instructions issued
         PAPI_TOT_INS:   Instructions completed
Derived:
         PAPI_FP_INS:    Floating point instructions
         PAPI_FP_OPS:    Floating point operations
         PAPI_LD_INS:    Load instructions
         PAPI_LST_INS:   Load/store instructions completed
         PAPI_SR_INS:    Store instructions
         PAPI_VEC_INS:   Vector/SIMD instructions

Abel


Abel Licon
915-867-4373
http://www.cis.udel.edu/~licon

"That's Life!"
-Frank Senatra

Re: [PerfSuite-users] Fatal Error

[Rick K. <rk...@nc...>]

Abel,

This is a pretty common issue that comes up from time to time with 
people installing for the first time.  Here is a prior posting to the 
perfsuite-users mailing list from 1/25/08 that should help resolve:

http://sourceforge.net/mailarchive/message.php?msg_name=Pine.LNX.4.64.0801250914180.18418%40osage.ncsa.uiuc.edu

Rick

Abel Licon wrote:
> I installed PAPI and perfsuite on a x86_64 box running suse10.1. PAPI  
> runs fine. I wanted to do some performance analysis on a molecular  
> dynamics simulation package, so I installed perfsuite as well.  
> However, when I try to run psrun, I get the following error.
>
> libpsrun fatal error: hardware event not available
>
> Here is the output of perfex -l and further down is the output of  
> perfex -i
>
> perfex -l
> CPU type Intel Pentium 4 Model 3
> One time-stamp counter available
> 18 performance counters available
> Overflow interrupts available
>
> Events Available:
> P4_TC_deliver_mode
> P4_BPU_fetch_request
> P4_ITLB_reference
> P4_memory_cancel
> P4_memory_complete
> P4_load_port_replay
> P4_store_port_replay
> P4_MOB_load_replay
> P4_page_walk_type
> P4_BSQ_cache_reference
> P4_IOQ_allocation
> P4_IOQ_active_entries
> P4_FSB_data_activity
> P4_BSQ_allocation
> P4_bsq_active_entries
> P4_SSE_input_assist
> P4_packed_SP_uop
> P4_packed_DP_uop
> P4_scalar_SP_uop
> P4_scalar_DP_uop
> P4_64bit_MMX_uop
> P4_128bit_MMX_uop
> P4_x87_FP_uop
> P4_x87_SIMD_moves_uop
> P4_TC_misc
> P4_global_power_events
> P4_tc_ms_xfer
> P4_uop_queue_writes
> P4_retired_mispred_branch_type
> P4_retired_branch_type
> P4_resource_stall
> P4_WC_Buffer
> P4_b2b_cycles
> P4_bnr
> P4_snoop
> P4_response
> P4_front_end_event
> P4_execution_event
> P4_replay_event
> P4_instr_retired
> P4_uops_retired
> P4_uop_type
> P4_branch_retired
> P4_mispred_branch_retired
> P4_x87_assist
> P4_machine_clear
> P4M3_instr_completed
>
> perfex -i
> PerfCtr Info:
> abi_version             0x05010501
> driver_version          2.6.25 DEBUG
> cpu_type                16 (Intel Pentium 4 Model 3)
> cpu_features            0x7 (rdpmc,rdtsc,pcint)
> cpu_khz                 2992615
> tsc_to_cpu_mult         1
> cpu_nrctrs              18
> cpus                    [0], total: 1
> cpus_forbidden          [], total: 0
>
> Thanks in advance for any help.
>
> -------------------------------------------------------------------------
> This SF.net email is sponsored by: Microsoft
> Defy all challenges. Microsoft(R) Visual Studio 2008.
> http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/
> _______________________________________________
> PerfSuite-users mailing list
> Per...@li...
> https://lists.sourceforge.net/lists/listinfo/perfsuite-users
>
>   

[PerfSuite-users] Fatal Error

[Abel L. <abe...@gm...>]

I installed PAPI and perfsuite on a x86_64 box running suse10.1. PAPI  
runs fine. I wanted to do some performance analysis on a molecular  
dynamics simulation package, so I installed perfsuite as well.  
However, when I try to run psrun, I get the following error.

libpsrun fatal error: hardware event not available

Here is the output of perfex -l and further down is the output of  
perfex -i

perfex -l
CPU type Intel Pentium 4 Model 3
One time-stamp counter available
18 performance counters available
Overflow interrupts available

Events Available:
P4_TC_deliver_mode
P4_BPU_fetch_request
P4_ITLB_reference
P4_memory_cancel
P4_memory_complete
P4_load_port_replay
P4_store_port_replay
P4_MOB_load_replay
P4_page_walk_type
P4_BSQ_cache_reference
P4_IOQ_allocation
P4_IOQ_active_entries
P4_FSB_data_activity
P4_BSQ_allocation
P4_bsq_active_entries
P4_SSE_input_assist
P4_packed_SP_uop
P4_packed_DP_uop
P4_scalar_SP_uop
P4_scalar_DP_uop
P4_64bit_MMX_uop
P4_128bit_MMX_uop
P4_x87_FP_uop
P4_x87_SIMD_moves_uop
P4_TC_misc
P4_global_power_events
P4_tc_ms_xfer
P4_uop_queue_writes
P4_retired_mispred_branch_type
P4_retired_branch_type
P4_resource_stall
P4_WC_Buffer
P4_b2b_cycles
P4_bnr
P4_snoop
P4_response
P4_front_end_event
P4_execution_event
P4_replay_event
P4_instr_retired
P4_uops_retired
P4_uop_type
P4_branch_retired
P4_mispred_branch_retired
P4_x87_assist
P4_machine_clear
P4M3_instr_completed

perfex -i
PerfCtr Info:
abi_version             0x05010501
driver_version          2.6.25 DEBUG
cpu_type                16 (Intel Pentium 4 Model 3)
cpu_features            0x7 (rdpmc,rdtsc,pcint)
cpu_khz                 2992615
tsc_to_cpu_mult         1
cpu_nrctrs              18
cpus                    [0], total: 1
cpus_forbidden          [], total: 0

Thanks in advance for any help.

Re: [PerfSuite-users] Native Events supported in Perfsuite?

[Rick K. <rk...@nc...>]

Chiaheng,

PerfSuite supports PAPI native events within its XML configuration 
files.  The difference between using PAPI presets and native events is 
that the XML attribute "type" should be set to "native", and the name of 
the event to be used should be given as the content of the element 
"ps_hwpc_event" rather than the value of the "name" attribute.

You can find an example of using native events in 
$PREFIX/share/perfsuite/xml/pshwpc/papi3_ipc_itanium2.xml, where PREFIX 
is the root of your PerfSuite installation.

Rick


> Subject:
> [PerfSuite-users] Native Events supported in Perfsuite?
> From:
> chiaheng <d94...@cs...>
> Date:
> Fri, 22 Feb 2008 21:15:49 +0800
>
> To:
> per...@li...
>
>
> Dear all,
>
>   This is Chiaheng, a graduate student who is new to Perfsuite tool.
>   I am now using Perfsuite to measure performance characteristics of 
> applications, however, i found that there are still some "native 
> events" exposed by PAPI library, which is not included in the xml file 
> along with perfsuite tool.
>   Would there be any high-level method to access those native events 
> via perfsuite, in particular, psrun command?
>
> thanks in advance,
> Chiaheng. 

[PerfSuite-users] PerfSuite developer position available

[Rick K. <rk...@nc...>]

Greetings -

Thanks to new funding support from the National Science Foundation, 
there is an opportunity to join in development of PerfSuite as described 
in the following employment posting from NCSA:

http://www.ncsa.uiuc.edu/AboutUs/Employment/11659.html

This is an exciting development that should help accelerate the 
evolution of PerfSuite greatly.  Please feel free to share this 
information with anyone who may be interested in helping develop this 
open source project (and yes, get paid for it!).  Note: job posting 
closes on March 14.

Rick

Re: [PerfSuite-users] Periodic measurement

[Rick K. <rk...@nc...>]

Sanjay,

An error code like that being returned from ps_hwpc_read() usually means 
that one or both of ps_hwpc_init()/ps_hwpc_start() have not been called 
before calling ps_hwpc_read(). I can't say if it's related to use of 
Charm++ (I assume you mean the software developed by UIUC's Parallel 
Programming Laboratory), but each process should call init, then start, 
then read.  In the case of a multithreaded program, only the main thread 
should call init but all threads should call start, read, etc.  However, 
because signal delivery is unpredictable with POSIX threads, I'd guess 
this would cause problems with the approach used in the "sampler" example.

Rick

Sanjay H A wrote:
> Hello Sir
>
> Progress about MPI : I am able to get output with MPI application.
>
> But one of my MPI application is using "charm++", On that if I insert 
> the routines
>
>
> The application runs for given sampling seconds, then inside the
> sampler subroutine, after calling ps_hwpc_read(), the program is 
> exiting. (ret value is not equal to success)
>
> The return value from ps_hwpc_read() is 1012.
>
> The error it is giving while exiting is: "calling sequence not allowed"
>
> Any suggestion on this. Is this because of charm++?
>
> Thanx
> Regards
>
> sanjay
>

Re: [PerfSuite-users] Periodic measurement

[Sanjay H A <san...@ly...>]

Hello Sir
<br>
<br>Progress about MPI : I am able to get output with MPI application.
<br>
<br>But one of my MPI application is using "charm++", On that if I insert the routines
<br>
<br>
<br>The application runs for given sampling seconds, then inside the
<br>sampler subroutine, after calling ps_hwpc_read(), the program is exiting. (ret value is not equal to success)
<br>
<br>The return value from ps_hwpc_read() is 1012.
<br>
<br>The error it is giving while exiting is: "calling sequence not allowed"
<br>
<br>Any suggestion on this. Is this because of charm++?
<br>
<br>Thanx
<br>Regards
<br>
<br>sanjay
<br><br><br><br><br><br><blockquote style='border-top: 1px solid rgb(204, 204, 204); margin: 0.8ex 0pt 0pt 0pt; padding-bottom: 1ex;'>
<br>---------[ Received Mail Content ]----------<br>
<br> <b>Subject : </b>Re: [PerfSuite-users] Periodic measurement<br>
<br> <b>Date : </b>Wed, 30 Jan 2008 11:10:04 -0600<br>
<br> <b>From : </b>Rick Kufrin &lt;rk...@nc...&gt;<br>
<br> <b>To : </b>Sanjay H A &lt;san...@ly...&gt;<br>
<br> <b>Cc : </b>per...@li...<br>
<br><br>
<br>Sanjay,
<br><br>
<br><br>The example programs like sampler that are distributed with PerfSuite 
<br><br>are not meant to be production-quality tools, but small cases that might 
<br><br>be useful for people to see how things work and demonstrate some basic 
<br><br>capabilities.  However, you might want to think about what's happening 
<br><br>underneath if you base something on an example like sampler.  It might 
<br><br>be helpful for you to take a look at the manual pages for setitimer() 
<br><br>and sigaction(), which are used in sampler as building blocks.
<br><br>
<br><br>Here are a few points that may be relevant:
<br><br>
<br><br>1. For performance data to be printed within sampler, the signal handler 
<br><br>that is installed must be triggered by delivery of the signal.  The 
<br><br>handler itself is the function "sampler".  As an aside, this is not 
<br><br>typically a good idea to be doing output from within a signal handler 
<br><br>but again this is just a toy example program.
<br><br>
<br><br>2. The sampler program sets interval timers using the ITIMER_VIRTUAL 
<br><br>time domain, which is decremented only when the process is executing.   
<br><br>If a process is sleeping, it's not accumulating any execution time, and 
<br><br>so this timer would likely be decrementing at a much slower rate.
<br><br>
<br><br>3.  The sampler program, in its use of sigaction, does not request that 
<br><br>system calls are restarted across signals (see the manual page for 
<br><br>sigaction()).
<br><br>
<br><br>I ran some tests and found that even with calls to printf() or sleep() 
<br><br>within the busyloop function, output from the signal handler will occur, 
<br><br>but not necessarily at the same rate one would expect when running the 
<br><br>example unchanged.  You might want to play around with it some more to 
<br><br>see if you can produce similar results.
<br><br>
<br><br>One other point to raise in case you make some changes is that PAPI, 
<br><br>when multiplexing, also uses signals to trigger the timeslicing of 
<br><br>counters (in most cases, I believe it uses SIGPROF and ITIMER_PROF).  
<br><br>That means some care has to be taken if changing signals and interval 
<br><br>timers within sampler.
<br><br>
<br><br>Not trying to be evasive here, but rather, trying to point you in a 
<br><br>different direction in terms of how you are thinking things are working 
<br><br>with the sampler example.
<br><br>
<br><br>Rick
<br><br>
<br><br>Sanjay H A wrote:
<br><br>> Hello Sir
<br><br>> I got the point. Then I will insert persuite routines in my application to measure performance periodically.
<br><br>> But I am afraid about the test I did with perfsuite sampler program.The thing I noticed is, inside busyloop(), if I insert printf statement or some sleep statement , the performance data will not 
<br><br>>  be printed on standard output periodically. Any solutions for that?
<br><br>>
<br><br>> Thanx
<br><br>> Regards
<br><br>> Sanjay
<br><br>>  
<br><br>>
<br><br>>
<br><br>>   
<br><br></BLOCKQUOTE>

Re: [PerfSuite-users] Periodic measurement

[Rick K. <rk...@nc...>]

Sanjay,

The example programs like sampler that are distributed with PerfSuite 
are not meant to be production-quality tools, but small cases that might 
be useful for people to see how things work and demonstrate some basic 
capabilities.  However, you might want to think about what's happening 
underneath if you base something on an example like sampler.  It might 
be helpful for you to take a look at the manual pages for setitimer() 
and sigaction(), which are used in sampler as building blocks.

Here are a few points that may be relevant:

1. For performance data to be printed within sampler, the signal handler 
that is installed must be triggered by delivery of the signal.  The 
handler itself is the function "sampler".  As an aside, this is not 
typically a good idea to be doing output from within a signal handler 
but again this is just a toy example program.

2. The sampler program sets interval timers using the ITIMER_VIRTUAL 
time domain, which is decremented only when the process is executing.   
If a process is sleeping, it's not accumulating any execution time, and 
so this timer would likely be decrementing at a much slower rate.

3.  The sampler program, in its use of sigaction, does not request that 
system calls are restarted across signals (see the manual page for 
sigaction()).

I ran some tests and found that even with calls to printf() or sleep() 
within the busyloop function, output from the signal handler will occur, 
but not necessarily at the same rate one would expect when running the 
example unchanged.  You might want to play around with it some more to 
see if you can produce similar results.

One other point to raise in case you make some changes is that PAPI, 
when multiplexing, also uses signals to trigger the timeslicing of 
counters (in most cases, I believe it uses SIGPROF and ITIMER_PROF).  
That means some care has to be taken if changing signals and interval 
timers within sampler.

Not trying to be evasive here, but rather, trying to point you in a 
different direction in terms of how you are thinking things are working 
with the sampler example.

Rick

Sanjay H A wrote:
> Hello Sir
> I got the point. Then I will insert persuite routines in my application to measure performance periodically.
> But I am afraid about the test I did with perfsuite sampler program.The thing I noticed is, inside busyloop(), if I insert printf statement or some sleep statement , the performance data will not 
>  be printed on standard output periodically. Any solutions for that?
>
> Thanx
> Regards
> Sanjay
>  
>
>
>   

Re: [PerfSuite-users] Periodic measurement

[Sanjay H A <san...@ly...>]

Hello Sir
I got the point. Then I will insert persuite routines in my application to measure performance periodically.
But I am afraid about the test I did with perfsuite sampler program.The thing I noticed is, inside busyloop(), if I insert printf statement or some sleep statement , the performance data will not 
 be printed on standard output periodically. Any solutions for that?

Thanx
Regards
Sanjay
 





---------[ Received Mail Content ]----------

 Subject : Re: [PerfSuite-users] Periodic measurement

 Date : Tue, 29 Jan 2008 11:12:05 -0600

 From : Rick Kufrin <rk...@nc...>

 To : Sanjay H A <san...@ly...>

 Cc : per...@li...


Sanjay,

PerfSuite uses a self-monitoring (first person) model, rather than one 
that monitors external processes/threads. I'm afraid the style of 
measurement you're describing is not likely to be available through 
PerfSuite. There may be other tools available out there that can do the 
sort of thing you're interested in, however. There are certainly tools 
that can "attach" to external processes, but the tricky part is 
obtaining the periodic sampling as the process runs.

Rick

Sanjay H A wrote:
> Hello Sir
>
> I tested sampler program, it worked.
>
> 1. One thing I noticed is, inside busyloop(), if I insert printf statement or some sleep statement , the performance data will not 
> be printed on standard output. May I know the reason?
>
> 2. If I want to run some external program instead of busy loop (Using system command). The performance data will not be printed . sample code given below
>
> int main(int argc, char **argv)
> {
> process_options(argc, argv);
>
> setup_pshwpc();
>
> setup_alarm(seconds);
>
> // busyloop();
> system("../loop 10000 >> temp "); // calling my program
> shutdown_pshwpc(write_results);
>
> exit(0);
> }
>
> What I assumed is, PID of the process will be different from the parent process (Thats why it is not printing the performance data) Is it true? Is there any solution for this type of problem.
>
> My point is I dont want to alter the code of my application. Instead I want to call my application from sampler program. I want to periodicaly measure the performance parameters !
> of my application
>
> Thanx
> Regards
>
> Sanjay
> This SF.net email is sponsored by: Microsoft
> Defy all challenges. Microsoft(R) Visual Studio 2008.
>
>
>
>
>
> ---------[ Received Mail Content ]----------
>
> Subject : Re: [PerfSuite-users] Periodic measurement
>
> Date : Sat, 26 Jan 2008 10:26:54 -0600 (CST)
>
> From : Rick Kufrin 
>
> To : Sanjay H A 
>
> Cc : per...@li...
>
>
> Sanjay,
>
> This type of periodic sampling is not supported directly by the psrun 
> command. It can be done via the PerfSuite API, however output format is 
> up to your own code. There is an example contained in the PerfSuite 
> distribution called "sampler" that shows how to set up interval timers 
> that trigger a read of the current performance data values. You can find 
> the example in the directory $PREFIX/share/perfsuite/examples/sampler
> (where $PREFIX is the value supplied to the configure option --prefix, or 
> /usr/local if nothing was supplied).
>
> Rick
>
> On Sat, 26 Jan 2008, Sanjay H A wrote:
>
> 
>> Hello Sir
>> Thanx for informative reply.
>> After commenting the events in the XML file 
>> that are not supported on my system, psrun worked.
>>
>> I want to measure 
>> performance events periodicaly for an long running application. Means If 
>> I have application runs for 1 hour, I want to measure these performace 
>> parameters for every 5 minutes.Is there any options in perfsuite? What I 
>> noticed is, if I run application with psrun, it creates xml file when 
>> the application finishes.
>>
>> Thanx again
>>
>> Regards
>> Sanjay
>>
>>
>>
>>
>> 
>
> -------------------------------------------------------------------------
> This SF.net email is sponsored by: Microsoft
> Defy all challenges. Microsoft(R) Visual Studio 2008.
> http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/
> _______________________________________________
> PerfSuite-users mailing list
> Per...@li...
> https://lists.sourceforge.net/lists/listinfo/perfsuite-users
>
>