You can think of a linked-list as a 1-nary tree. Interesting to think about fine-grain locking performance on various k-nary trees.
As k gets larger, how would fine-grained locking performance change?
temmie
@0xc0ffee I guess I would expect the advantage of fine-grained locking relative to coarse locks to increase with k on k-nary trees, since each locked node blocks more potential paths down the tree.
1pct
Considering the coarse locks and fine locks for hash-table, why do they have the same execution times when there is 1 processor rather than the fine locks have longer execution times due to lock overhead like the second graph.
You can think of a linked-list as a 1-nary tree. Interesting to think about fine-grain locking performance on various k-nary trees.
As k gets larger, how would fine-grained locking performance change?
@0xc0ffee I guess I would expect the advantage of fine-grained locking relative to coarse locks to increase with k on k-nary trees, since each locked node blocks more potential paths down the tree.
Considering the coarse locks and fine locks for hash-table, why do they have the same execution times when there is 1 processor rather than the fine locks have longer execution times due to lock overhead like the second graph.