[
Date Prev][
Date Next][
Thread Prev][
Thread Next][
Date Index][
Thread Index]
[
List Home]
[henshin-dev] Optimizations
|
Hi,
I implemented a number of optimizations for the state space tools. Most
importantly, the hash code function for EMF models is now a 'real' graph
hash function. This was the main bottleneck up to now; the performance
gain from this change alone is tremendous for graph-based equality
checking. Additionally, the actual graph isomorphy checking is done
using the matching engine now, and for the state space generation a
static analysis is used to determine a partial match in advance before
doing the isomorphy checking.
There are still a couple of bugs that need to be fixed and I think that
the performance can be further improved. My guess is that it is still
slower than GROOVE but we are getting better. Once the bugs are all
fixed, I will set up some tests which we can also use for benchmarks
then. At some point we can think about writing a technical report about
the matching algorithms that we use in Henshin, because I think we
already gathered a number of interesting optimizations which are worth
to publish.
Cheers,
Christian