alarsyo/spot
1
0
Fork 0
Code Issues Pull requests Projects Releases Wiki Activity

bench/stutter: add a "user" benchmark

Browse source 
* bench/stutter/user.sh: New bench.
* bench/stutter/Makefile.am: Add it.
* bench/stutter/README: Mention it.
* bench/stutter/stutter_bench.sh: Run it.
This commit is contained in:
Alexandre Duret-Lutz 2015-02-13 16:58:42 +01:00
parent 204af40b09
commit e900488e13
4 changed files with 47 additions and 15 deletions
Download patch file Download diff file Expand all files Collapse all files

26
bench/stutter/README
View file

@ -1,20 +1,22 @@
This benchmark measures the performance of different algorithms to
check if property (expressed as a formula or as a deterministic TGBA)
is stutter-invariant. When the benchmark is run on formulas, the
translation time is not included in the measured time.
check if a property is stutter-invariant. When the benchmark is run
on formulas, the translation time is not included in the measured
time.
To reproduce the benchmark is to run
To reproduce the benchmark run
% ./stutter_bench.sh -j8
to create bench_formulas.csv and bench_randgraph.csv.
(Adjust -j8 to the number of cores you have.)
Then explore these data the provided ipython notebook
This should create three files:
% ipython notebook --pylab=inline stutter.ipynb
The time in bench_formulas.csv is reported in microseconds, while the
time in bench_randgraph.csv is in seconds.
- bench_formulas.csv
running stutter checks on LTL formulas, but without including
LTL translation time.
- bench_randgraph.csv
unning stutter checks on deterministic TGBA, without including
complementation time.
- ltl-user-bench.csv
running stutter checks on LTL formulas, from the user point
of view, i.e., timing the whole "ltlfilt" run.