Sylvain Bougerel - 2016-01-03
  • Description has changed:

Diff:

--- old
+++ new
@@ -1 +1 @@
-This change translate only in slightly better performance when using a Static_rank in a container. However for Dynamic_rank, it makes a big difference, since the operator% cannot be optimized as much, and bookkeeping of dim requires more operator% operations than bookkeeping of depth.
+This change translate only in slightly better performance when using a Static_rank in a container. However for Dynamic_rank, it makes a big difference, since the operator% cannot be optimized as much, and bookkeeping of dim requires more operator% operations than bookkeeping of depth. It has to be tested independantly from each iterators. Some iterators, like neighbor_iterators, benefit from the "dim" rather than the "depth" strategy.