[cpp-threads] memory model
Alexander Terekhov
alexander.terekhov at gmail.com
Fri May 6 01:02:44 BST 2005
On 5/5/05, Boehm, Hans <hans.boehm at hp.com> wrote:
[... ddacq and ccacq ...]
> I understand the motivation. But I'm not convinced it's worth
> the cost, at least for the control dependent version. Certainly
> if it affects optimization of code that doesn't obviously use
> these primitives, it isn't.
I don't know. Of course fake dependencies are not meant to
affect optimizations. As for cc <-> dd transformations,
all power to the optimizer as long as it doesn't break the logic
expressed in source code. Usual "as if'"business, oder?
regards,
alexander.
More information about the cpp-threads
mailing list