search for: bddbddb

...as interested to see whether Whaley's algorithm really works > particularly, given its simplicity and the claims made in the papers. I have implemented a tool based on the Phoenix compiler framework at MSR, which can dump relations as BDD from C/C++ code and feed them into John Whaley's bddbddb for further analysis. As Dan said, currently the performance is not good enough for C/C++ code (due to more complex relations, BDD variable orders, etc.). I am still trying several optimizations. John also told that it took quite a lot of effort to get the Java version to scale. I am interested...

> No > I've got a BDD based version of andersen's around already (uncommitted). > It's slower in time (by a factor of 8 or so), but very slightly (1-2 > megabytes of memory) more memory efficient on some cases. Nowadays, > the bitmap version is actually more memory efficient than the BDD > version, even if you turn off BDD caching, etc. > > The thing you

...Whaley's algorithm really works > > particularly, given its simplicity and the claims made in the papers. > > I have implemented a tool based on the Phoenix compiler framework at > MSR, which can dump relations as BDD from C/C++ code and feed them > into John Whaley's bddbddb for further analysis. As Dan said, > currently the performance is not good enough for C/C++ code (due to > more complex relations, BDD variable orders, etc.). I am still trying > several optimizations. John also told that it took quite a lot of > effort to get the Java version to...