> As far as I know, the cost for cache coherency comes when core-to-core 
> transfer is required.

Can we expect thread-local storage to be faster than shared memory ?

Reply via email to