On 2016-05-21 16:06, Piotr Szturmaj wrote:
https://www.cs.bgu.ac.il/~hendlerd/papers/flat-combining.pdf [2010]
"Traditional data structure designs, whether lock-based or
lock-free, provide parallelism via fine grained synchronization
among threads.
We introduce a new synchronization paradigm based on
coarse locking, which we call flat combining. The cost of
synchronization in flat combining is so low, that having a
single thread holding a lock perform the combined access
requests of all others, delivers, up to a certain non-negligible
concurrency level, better performance than the most effective
parallel finely synchronized implementations. We use
flat-combining to devise, among other structures, new linearizable
stack, queue, and priority queue algorithms that
greatly outperform all prior algorithms."
Another relevant article:
https://software.intel.com/en-us/blogs/2013/02/22/combineraggregator-synchronization-primitive
Combiner is an alternative to mutex, and basically takes critical
sections from many threads, combines and executes them on one thread
(the one that currently has lock), leading to non-negligible boost in
performance due to cache locality. If there are no combinining
opportunities then it behaves exactly like a mutex.
I'm leaving it here in case someone would like to implement this pattern
in D language :)
