3 ms·
> So yes, that "++*barrier" operation is read-modify-write atomically AND with memory-barriers in sequential-consistency style The quoted material does not des
by MaxGanzII 5y ago
> So yes, that "++*barrier" operation is read-modify-write atomically AND with memory-barriers in sequential-consistency style
The quoted material does not describe this assertion.
- dragontamer 5y ago> Built-in increment and decrement operators and compound assignment are read-modify-write atomic operations with total sequentially consistent ordering (as if using memory_order_seq_cst). If less strict synchronization semantics are desired, the standard library functions may be used instead.