[cpp-threads] High-level vs. low-level

[Alexander Terekhov]

On 7/13/06, Peter Dimov <pdimov at mmltd.net> wrote:
[...]
> You have to pick one or the other :-) The current model behind the
> high-level loads and stores is (IIUC) ld.acq+st.rel, which has been proven
> to result in close to optimal performance.
>
> I'm not sure that a sequentially consistent model can achieve the above
> "within 5%" performance goal. On x86, for example, all stores would need to
> be fully locked (lock xchg).

and all loads would need to be fully lock cmpxchg 42. (To provide
SC-ish illusion of remote write atomicity.)

regards,
alexander.