Development/Score-P: Difference between revisions
No edit summary |
|||
| Line 13: | Line 13: | ||
|- |
|- |
||
|Citing |
|Citing |
||
| A. Knüpfer et al, [http://link.springer.com/chapter/10.1007%2F978-3-642-31476-6_7 Score-P: A Joint Performance Measurement Run-Time Infrastructure for Periscope, Scalasca, TAU, and Vampir], in Tools for High Performance Computing 2011, pages 79-91, Springer. |
|||
| n/a |
|||
|- |
|- |
||
| Links |
| Links |
||
| Line 88: | Line 88: | ||
= Documentation = |
= Documentation = |
||
For '''online documentation''' see the links section in the summary table at the top of this |
For '''online documentation''' see the links section in the summary table at the top of this page. |
||
For '''local documentation''' please check the manual below the <kbd>$SCOREP_DOC_DIR</kbd> directory using a browser to: |
|||
<pre>$ man gdb</pre> |
|||
<pre> |
|||
firefox $SCOREP_DOC_DIR/scorep/html/ |
|||
</pre> |
|||
The man pages added are for the (sub-project) PAPI, i.e. the CPU performance counters. |
|||
<br> |
<br> |
||
Revision as of 20:47, 5 November 2020
| Description | Content |
|---|---|
| module load | devel/scorep |
| Availability | bwUniCluster | BwForCluster_Chemistry |
| License | BSD 3-clause License |
| Citing | A. Knüpfer et al, Score-P: A Joint Performance Measurement Run-Time Infrastructure for Periscope, Scalasca, TAU, and Vampir, in Tools for High Performance Computing 2011, pages 79-91, Springer. |
| Links | Homepage | Documentation | News Mailinglists | Bug Reports |
| Graphical Interface | No |
| Included modules | - |
Introduction
The Score-P Project is a profiling and tracing library with supportive tools developed mainly by by the [VI-HPS] project.
Versions and Availability
A list of versions currently available on all bwHPC Clusters can be obtained from the
Cluster Information System CIS
{{#widget:Iframe |url=https://cis-hpc.uni-konstanz.de/prod.cis/bwUniCluster/devel/scorep |width=99% |height=120 }} On the command line interface of any bwHPC cluster you'll get a list of available versions by using the command module avail devel/scorep. Please note that Score-P installation may be hidden in the sense that it's dependent on the MPI, which in turn is dependent on the compiler. So while there are a multitude of compiler version's only some may feature an MPI, which offers tracing using Score-P. Please use module --ignore-cache spider to evaluate all possible installations.
$ : bwUniCluster 2.0 $ module load compiler/gnu/10.2 mpi/openmpi ; module avail devel/score ------------------------ /opt/bwhpc/common/modulefiles/Core ------------------------- devel/scorep/6.0
Loading
Score-P may be loaded with a specific and MPI:
$ module load compiler/gnu/10.2 mpi/openmpi devel/scorep
Check the installation of this Score-P instance using scorep-info here with some important output:
$ scorep-info config-summary
...
Loaded modules:
module load compiler/gnu/10.2 \
mpi/openmpi/4.0 \
devel/cuda/10.2
...
PAPI support: yes
OpenCL support: yes
libOpenCL support: yes, using -lOpenCL
...
CUDA support: yes
libcudart support: yes, using -I/opt/bwhpc/common/devel/cuda/10.2/include -L/opt/bwhpc/common/devel/cuda/10.2/lib64 -R/opt/bwhpc/common/devel/cuda/10.2/lib64 -lcudart
libcuda support: yes, using -L/opt/bwhpc/common/devel/cuda/10.2/targets/x86_64-linux/lib/stubs/ -R/opt/bwhpc/common/devel/cuda/10.2/targets/x86_64-linux/lib/stubs/ -lcuda
libcupti support: yes, using -I/opt/bwhpc/common/devel/cuda/10.2/include -I/opt/bwhpc/common/devel/cuda/10.2/extras/CUPTI/include -L/opt/bwhpc/common/devel/cuda/10.2/extras/CUPTI/lib64 -R/opt/bwhpc/common/devel/cuda/10.2/extras/CUPTI/lib64 -lcupti
CUPTI async support: yes
CUDA version >= 6.0: yes
...
OpenMP support: yes, using -fopenmp
OpenMP pomp_tpd: yes
OpenMP ancestry: yes
Documentation
For online documentation see the links section in the summary table at the top of this page. For local documentation please check the manual below the $SCOREP_DOC_DIR directory using a browser to:
firefox $SCOREP_DOC_DIR/scorep/html/
The man pages added are for the (sub-project) PAPI, i.e. the CPU performance counters.