On Sun, 2004-04-11 at 19:03, Dr. David Alan Gilbert wrote:
> Be careful in making judgements about the inlining behaviour based
> on sizes on one architecture.  The size of these functions
> is likely to be radically different on different architectures, and
> the break off point of cache/RAM usage v branch cost is also likely
> to be entirely different.

this is why such a choice most likely should be up to gcc.....
and it's getting a lot better at that nowadays.