Changes between Version 30 and Version 31 of pgraph

Timestamp:

Jan 14, 2010, 11:24:10 PM (14 years ago)

Author:

asturzu

Comment:

--

pgraph

@@ -36 +36 @@
 The results we obtained are in the powerpoint presentation located [[BR]] in the svn repository
 in the folder named Presentation.
+For the Dijkstra and Prim parallel algorithms the profiler showed that the bottleneck [[BR]]
+was produced mainly by the MPI_Allreduce function.
+[[BR]] For the Bellman-Ford algorithm, the poor results were caused clearly by
+the thread creation and destruction overhead in OpenMP.
+
 
 == References ==