'false sharing' == two items next to another in memory are accessed by different thread, as they are closed in memory they are then in the same cache line which

'false sharing' == two items next to another in memory are accessed by different thread, as they are closed in memory they are then in the same cache line which is 'shared' between two CPUs: read only is okay but read&write is a performance killer Here it's a cache (associativity) conflict, as gugagore said it can impacts performance even in read only situation.. And yes, you're describing a way to avoid the conflicts.