On 05/30/2016 03:49 PM, Peter Zijlstra wrote: >> The same thing can happen within a Task if you take my story from above >> and replace task with thread. Completely understood. > > Right; so I don't see the point of PREALLOCATE_HASH to cater for RT > workloads if it still doesn't guarantee anything, esp. if the failure > case is silent and obscure.
So what do you suggest? Adding trace points in order to learn about possible collisions or using tickets (on top of this) to guarantee being collision free? Note: this as it, is already a win on NUMA boxes since the memory is not referenced cross node. Sebastian