- Infos im HLRS Wiki sind nicht rechtsverbindlich und ohne Gewähr -
- Information contained in the HLRS Wiki is not legally binding and HLRS is not responsible for any damages that might result from its use -
Performance Analysis Tools: Difference between revisions
From HLRS Platforms
Jump to navigationJump to search
No edit summary |
No edit summary |
||
| Line 3: | Line 3: | ||
There are several tools to analyse the performance of your application. Some are pre-installed on our systems and ready to use. Please see the documentation to get more details of every tool. | There are several tools to analyse the performance of your application. Some are pre-installed on our systems and ready to use. Please see the documentation to get more details of every tool. | ||
Please make sure to always set ''MPI_SHEPHERD=true'' when using performance analysis tools. Otherwise you might crash the nodes you are | Please make sure to always set ''MPI_SHEPHERD=true'' when using performance analysis tools. Otherwise you might crash the nodes you are working on. | ||
{| class="wikitable" | {| class="wikitable" | ||
Revision as of 17:34, 27 December 2021
Note: This page is in progress.
There are several tools to analyse the performance of your application. Some are pre-installed on our systems and ready to use. Please see the documentation to get more details of every tool.
Please make sure to always set MPI_SHEPHERD=true when using performance analysis tools. Otherwise you might crash the nodes you are working on.
| Tool | Purpose | Available on |
|---|---|---|
| Advisor | threading assistant | Hawk |
| Darshan | tracing IO performance of mpi programms | Hawk |
| Extrae | Hawk | |
| Kojak | ||
| LIKWID | ||
| OPT | ||
| Paraver | ||
| Scalasca | ||
| Score-P | tracing and sampling of MPI and Open MP parallel applications | Hawk, Vulcan |
| Threadspotter | ||
| Trace_Analyzer_and_Collector | ||
| Vampir | ||
| VTune |