| introduced more memory friendly edgeiditerator, all tests passing |
|
More...
|
over 12 years ago
|
| Revert "moved edges stuff into separate class" |
|
More...
|
over 12 years ago
|
| moved edges stuff into separate class |
|
More...
|
over 12 years ago
|
| skip sleep before dijkstra |
|
More...
|
over 12 years ago
|
| avoid filling refToEdges with -1 and ensure that edgepointer start with pos values; removed priorities; moved deleted nodes to more memory efficient open bitset |
|
More...
|
over 12 years ago
|
| minor refactoring |
|
More...
|
over 12 years ago
|
| fixed incompatibility to perf prj |
|
More...
|
over 12 years ago
|
| minor cleanup; increase factor for memgraph too |
|
More...
|
over 12 years ago
|
| moved again to embedded distances array as length was calculated too big (should be pointer/LEN_EDGE, was pointer) |
|
More...
|
over 12 years ago
|
| removed locks and using separate distance array to give a minor speed up for now |
|
More...
|
over 12 years ago
|
| added some more locks; without all locks our new graph is only 10% slower then in-memory. probably due to float conversion |
|
More...
|
over 12 years ago
|
| tried new storage. it is as slow as mmap graph |
|
More...
|
over 12 years ago
|
| ran findbugs |
|
More...
|
over 12 years ago
|
| minor readme |
|
More...
|
over 12 years ago
|
| made all tests passing |
|
More...
|
over 12 years ago
|
| refactored 'location' into 'node' of Graph interface |
|
More...
|
over 12 years ago
|
| added new in-memory graph structure with some nice features |
|
More...
|
over 12 years ago
|
| added double to bitutil |
|
More...
|
over 12 years ago
|
| when using approx distance calculation quadtree search is about 30% faster |
|
More...
|
over 12 years ago
|
| fixing some minor glitches; created approxcalcdistance to be more pluggable |
|
More...
|
over 12 years ago
|
| implemented the same for tinkergraph; even worse than neo4j; memory consumption a way too much (>1g for our 1mio nodes example) |
|
More...
|
over 12 years ago
|
| commented tests which need a unidirectional difference and thus will fail for neo4j |
|
More...
|
over 12 years ago
|
| minor perf improvement for neo4j |
|
More...
|
over 12 years ago
|
| finally get the same results for dijkstra with neo4j too but perf is still 8 times slower |
|
More...
|
over 12 years ago
|
| avoid id index and convert between neo4j and our id |
|
More...
|
over 12 years ago
|
| more improvements but still hellish slow and dijkstra does not work |
|
More...
|
over 12 years ago
|
| first successfully neo4j import |
|
More...
|
over 12 years ago
|
| furhter minor refactoring |
|
More...
|
over 12 years ago
|
| additional refactoring to try out a storage again for neo4j |
|
More...
|
over 12 years ago
|
| minor refactoring for perf-comparison |
|
More...
|
over 12 years ago
|
