I'm exhuming this rather old thread because I came across a related thread in the boost mailing list https://groups.google.com/forum/#!topic/boost-list/7nCzvkkEXCk .
If I get right, the problem is the same that I'm dealing with. The initialization of the distance / predecessor maps takes more time than the actual Dijkstra search because a very small region of the graph is visited. I guess that it is particularly true when the distance / predecessor maps do not fit in CPU cache.
Could you go through the proposed solution? If you confirm that the solution is suitable, I'll implement it in the bfs / dijkstra visitors.
-- View this message in context: http://main-discussion-list-for-the-graph-tool-project.982480.n3.nabble.com/... Sent from the Main discussion list for the graph-tool project mailing list archive at Nabble.com.