Subj : Re: Memory Barriers, Compiler Optimizations, etc.
To   : comp.programming.threads
From : Neill Clift [MSFT]
Date : Sat Feb 05 2005 03:28 pm

"Alexander Terekhov" <terekhov@web.de> wrote in message 
news:420506AC.4D92C384@web.de...
>
> Same basis as with respect to preservation of sequence of std IO
> calls to dev/null (with destination known in advance so to speak).
> The thing is that C/C++ volatile abuse isn't really "observable
> behavior" under single-threaded C/C++ standard(s), you know.
>

In section 6.5.3 of the ANSI C standard as a footnote it says:

'A volatile declaration may be used to describe an object
corresponding to a memory-mapped input/output port...'

So compiler reordering would be visable via it's effect
on a hardware device.
I got this by looking in Mr Schildt's book :-)
Neill.

.