Determining exactly how well an MPI application is performing on current HPC systems is a challenging task. Analysis of the cputime, system time and IO time of a serial application can provide basic performance information but for a parallel application, the (wasted) time spent waiting on communication is not visible from “outside the application”. MPI performance analysis tools provide insight into this “internal” computation versus communication behaviour and, as a result, understanding of the application’s parallel performance. They can reveal potential issues such as load imbalance, synchronization contentions and much more. As well as pointing out the limitations of an MPI application, access to this profiling information can assist user in optimizing the application to achieve greater scalability.
MPI performance analysis is normally performed at two levels. The first level is called MPI summary profiling or simply MPI profiling, which aggregates statistics at run time and provides performance overview of the whole job execution. The second level, called MPI tracing, collects the MPI event history of an application execution and provides fine grained information for each MPI function call (every message passed) along the execution timeline.
This document describes how to use MPI performance analysis tools including profilers and tracers which are available on NCI NF compute systems. For further help with using MPI profilers and tracers, please send email to help@nci.org.au .
Contents
A MPI profiler aggregates “whole run” statistics at run time — e.g., total amount of time spent in MPI, total number of messages or bytes sent, etc. As this information is available on a per-rank basis, issues such as load imbalance are exposed.
Typically the overhead of collecting this summary profiling data is very low (~1%) and the volume of profiling data collected is also very low. During runtime, information collection is local to each process and simply involves updating counters each time an MPI call is made. The profiling library only invokes communication during report generation, typically at the end of the run, to merge results from all of the tasks into one output file. As a result, it is feasible to include the use of an MPI profiler in all production runs.
Note that (currently) no profiling information will be produced if the execution does not complete normally (i.e. does not call MPI_Finalize()).
On NCI NF compute systems, two different lightweight MPI profilers are installed. They are IPM and mpiP. Both of these tools require minimal actions to invoke – we recommend that you use them regularly. Note that their use is only applicable to Open MPI applications.
IPM is a widely-used MPI profiler with many sites using it by default for every MPI job. The level of detail is selectable at runtime and presented through a variety of text and web reports. IPM can also provide information from the CPU HW counters giving basic CPU as well as MPI statistics.
For version 1.6.5 or less, please use ipm/0.983-nci.For version 1.7 or more, please use ipm/2.0.2. Please note this is different to NCI version and we are working on our own version for 2.0.2.
Using IPM does not require code recompilation. Instead, LD_PRELOAD is used to dynamically load the IPM library (libipm.so) as a wrapper to the MPI runtime.
A simple PBS job script using the IPM profiler with an MPI executable (prog.exe) is as shown below:
{{{#!/bin/bash
mpirun prog.exe > output
}}} Currently, IPM is available for openmpi version 1.4.1 and above.
User can also define the IPM log directory and log file name via setting following environment variables in the PBS job script before mpirun, an good example of IPM log directory and file name are:
export IPM_LOGDIR=/short/$PROJECT/ipm_logs export IPM_LOGFILE=$PBS_JOBID.$USER.$PROJECT.`date +%s`
NOTE: For some applications, defining IPM_LOGDIR in .bashrc or .cshrc is compulsory to successfully generate IPM profile data file. These applications usually have following features:
different MPI ranks working in different directories,
PBS job scripts are automatically generated by some sense,
more to be added …
By default IPM produces a summary of the performance information for the application on stdout. IPM also generates an XML data file which will be named something like (if user hasn’t define IPM_LOGFILE envorinment variable):
your_username.1231369287.321103.0 eg. jxc900.1231369287.321103.0
The XML data file can be used to generate a graphical webpage in one of two ways.
To visualize the IPM XML data on Raijin, you need to login to Raijin with an X display, eg. using ssh -X or ssh -Y, or with VNC. The detailed sample instruction on Raijin is listed below.
$ ssh -X raijin $ module load openmpi $ module load ipm $ ipm_view IPM_XML_file
Alternatively, the IPM XML data file can be parsed to HTML format. The detailed sample instruction on raijin are:
$ ssh raijin $ module load openmpi $ module load ipm $ module load ploticus $ ipm_parse -html <IPM_XML_file>
The ipm_parse command will generate a directory containing parsed IPM profile data with graphs. The directory will be named something like:
a.out_1_your_username.1231369287.321103.0_ipm_${jobid}
You can secure copy the directory to your local disk. A sample instruction is as follows.
$ scp -r a.out_1_your_username.1231369287.321103.0_ipm_${jobid} user@your_local_machine:path_to_store/
Then you can view it with your favourite web browser, e.g firefox, on your desktop:
$ firefox path_to_store/a.out_1_your_username.1231369287.321103.0_ipm_${jobid}/index.html
IPM can be integrated with hardware performance counters and profile useful information such as GFlops, cache misses, etc. PAPI is used for this purpose. Currently, IPM-HPM are only avaialble for OpenMPI version 1.4.3.
To usage IPM with HPM, please do following:
module load ipm/0.983-hpm
A sample PBS job script will be similar to following:
#!/bin/bash #PBS -l ncpus=2 module load openmpi module load ipm/0.983-cache export IPM_HPM=PAPI_FP_OPS,PAPI_TOT_INS,PAPI_TOT_CYC,PAPI_L2_TCA mpirun prog.exe > output
The environment variable IPM_HPM can be selected from following predefined PAPI event sets for Raijin.
For Raijin:
1. PAPI_FP_OPS,PAPI_TOT_INS,PAPI_TOT_CYC,PAPI_L2_TCA 2. PAPI_FP_OPS,PAPI_TOT_INS,PAPI_TOT_CYC,PAPI_L2_TCA,PAPI_L2_TCM 3. PAPI_TOT_INS,PAPI_TOT_CYC,PAPI_L2_STM,PAPI_L2_LDM,PAPI_L2_TCM,PAPI_L2_DCM 4. PAPI_TOT_CYC,PAPI_TOT_INS,PAPI_L3_TCM,PAPI_L3_LDM,PAPI_SR_INS,PAPI_LD_INS 5. MEM_LOAD_RETIRED:OTHER_CORE_L2_HIT_HITM,MEM_UNCORE_RETIRED:OTHER_CORE_L2_HITM 6. PAPI_FP_OPS,PAPI_FP_INS,PAPI_DP_OPS,PAPI_VEC_DP 7. PAPI_FP_OPS,PAPI_FP_INS,PAPI_SP_OPS,PAPI_VEC_SP 8. PAPI_FP_OPS,PAPI_RES_STL,PAPI_TOT_CYC,PAPI_TOT_INS 9. PAPI_FP_OPS,PAPI_TLB_DM,PAPI_TLB_IM 10. PAPI_L1_DCA,PAPI_L2_DCM
You can also customize your own PAPI eventsets for IPM_HPM environment variable via using papi_event_chooser command.
$ module load papi $ papi_event_chooser -help
Both PAPI pre-define events, and native hardware events can be used.
setenv IPM_ROUNDED yes
This feature significantly reduces the memory usage and postprocessing time during both runtime and parsing time.
setenv IPM_CHECKPOINT yes setenv IPM_CHKPT_INTERVAL 3600
This feature allow IPM to checkpoint profile status in $IPM_CHKPT_INTERVAL seconds intervals. The checkpointed profile data will be stored in a.out_1_your_username.1231369287.321103.0_ipm_${jobid}.${rank} files.
These files can be merged (using cat command) into a single xml file, and parsed with ipm_parse.
mpiP is a lightweight profiling library for MPI applications. In addition to the MPI summary profiling provided by IPM, mpiP can provide “call site” statistics showing which calls in the code are dominating MPI execution time.
Using mpiP does not require code recompilation. Instead, LD_PRELOAD is used to dynamically load the mpiP library (libmpiP.so) as a wrapper to the MPI runtime.
A sample PBS job script using mpiP profiler with MPI executable (prog.exe) is as shown below:
#!/bin/bash #PBS -l ncpus=2 module load openmpi/1.3.3 module load mpiP mpirun prog.exe > output
Currently, mpiP is available for openmpi version 1.3.3 and above.
The mpiP profiler generates a text based output file named something like:
prog.exe.${np}.25972.1.mpiP
To visualize the mpiP profile data on Raijin, you need to login to Raijin with a X display, eg. using ssh -X or ssh -Y, or with VNC. The detailed sample instruction on raijin is listed below.
$ ssh -X raijin $ module load openmpi $ module load mpiP $ mpipview prog.exe.${np}.XXXXX.1.mpiP
mpipview is able to map MPI callsites in the profile data to source code. This requires the MPI program to be compiled with -g option and linked with libunwind, as follows.
$ module load openmpi $ module load mpiP $ mpicc -g -o prog.exe prog.c -lmpiP -lm -lbfd -liberty -lunwind
Due to MPI profilers only profile for MPI function calls, it is not sufficient to reveal other details of the application. To get a better knowledge of users program, for example:
it is necessary to use a general purpose profiler.
Please refer General profiling for the details of the general profilers available at NF compute systems.
An MPI tracer collects an event history. It is common to display such event history on a timeline display. Tracing data can provide much interesting detail, but data volumes are large and the overhead of collection may be non-trivial. Often the collection of traces has to be limited in both duration and number of cpus to be feasible. The use of MPI tracing is strongly encouraged during the development or tuning of parallel applications but should not be used in production runs.
In raijin:
In script of PBS,
module load openmpi/1.6.5-mlx module load darshan darshan-mk-log-dirs.pl //create the log directories for specific users with darshan-mk-log-dirs.pl /apps/openmpi/1.6.5-mlx/bin/mpirun ..
After running, generate pdf summary report from logs.
darshan-job-summary.pl /logdir/<JobID>***.darshan.gz
The NCI NF is investigating install the following MPI tracing tools:
More information will be provided shortly.
The Allinea DDT&MAP and its performance report products are installed on Raijin for evaluation with a trial license.
A brief homemade Usage Guide for Raijin (requested by Muhammad):
cp /apps/allinea/4.2/tools/templates/pbs.qtf ~ edit pbs.qtf to be suitable for raijin like adding #PBS -P z00 , #PBS -l mem 30GB
Implementation: choose openmpi with ticking “override default mpirun path” and using full path like “/apps/openmpi/1.6.5/bin/mpirun” Make sure you tick “Memory Debugging” for DDT
Replace your mpirun command line as perf-report /apps/openmpi/1.*/bin/mpirun .. (Note: full path of mpirun must be used)
More on usage can be found in raijin:/apps/allinea/4.2/tools/doc/userguide.pdf