Nonportable code in C++ support

[Matthew Woehlke]

Kuba Ober wrote:
> Are those tight loops realistic?

No, but I don't see where either algorithm would have noticeably 
different memory access characteristics, so it shouldn't matter. In both 
instances, you have to fetch the pre-hashed pointer.

Honestly? I doubt that the difference would be noticeable.

It would be interesting to write a "full" test, but I'm not sure I'd 
trust anything outside of an actual application (and then, how do you 
test performance? :-) ), as a realistic allocation pattern is needed.

-- 
Matthew
No animal should ever jump on the dining room furniture unless 
absolutely certain he can hold his own in conversation -- Fran Lebowitz