Commit 1e2ff0fa authored by Dominic Etienne Charrier's avatar Dominic Etienne Charrier
Browse files

Add some more advanced tableslicer/speedupcalculator usage patterns.

parent 5da2498a
......@@ -532,4 +532,23 @@ optional arguments:
Output file
\end{code}
\paragraph{Computing speedups if table contains different data sets}
If you use \texttt{sweep.py}, you often pipe output from
multiple different runs into the same results folder. Parsing these
outputs will then create a big CSV table containing data sets for multiple orders, mesh sizes
etc. In this case, you cannot use the speedup calculator tool directly as
it will take the first column as reference value.
Here, a combination of the tableslicer and the speedupcalculator tool can do the trick.
Assume your CSV table contains measurements for multiple polynomial orders and you
want to compute the speedup for a specific one.
In this case you can run, e.g.:
\begin{code}
./tableslicer.py <mypath>/<myproject>-timestep-times.csv --cols order cores realtime_min --filter order=5 | ./speedupcalculator.py
\end{code}
Bash further allows writing for loops. Therefore, you can compute speedups for all considered orders (3,5,7 in our example)
quite easily:
\begin{code}
for o in 3 5 7; do ./tableslicer.py <mypath>/<myproject>-timestep-times.csv --cols order cores realtime_min --filter order=$o | ./speedupcalculator.py; done
\end{code}
Markdown is supported
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment