This is the mail archive of the libstdc++@gcc.gnu.org mailing list for the libstdc++ project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: binary_search, lower_bound and upper_bound inlining


Thanks for answering,

> >Is there something that I can do, other than editing the stl_algo.h file, in order to persuade
> the
> >compiler to inline the binary_search, lower_bound and upper_bound algorithms? (even -O3 seems
> not
> >good enough). Any chance for these algorithms to be inlined in the future GCC releases?
> >  
> >
> we can definitely mark inline binary_search, very tiny, but frankly, for 
> general use lower_bound and upper_bound seem too big and we risk a bad 
> code bloat. 

I suspected so :( That's life anyway...

> Can you measure a small improvement for binary_search alone?

By studding my measurements I suspect that binary_search is automatically inlined in -O3 (probably
 because it is small). Anyway the "big fish" is lower_bound, binary_search is just more convenient
sometimes.  
 
> 
> In general, anyway, you should refer to the gcc manual, there are 
> various knobs you can turn in order to change the behavior of the 
> inliner (-finline-limit=...)

I found nothing appropriate in the manual. (-finline-limit works for functions that are explicitly
marked as inline)

Thank you very much,
-Jim



	

	
		
___________________________________________________________ 
Χρησιμοποιείτε Yahoo!; 
Βαρεθήκατε τα ενοχλητικά μηνύματα (spam); Το Yahoo! Mail 
διαθέτει την καλύτερη δυνατή προστασία κατά των ενοχλητικών 
μηνυμάτων http://login.yahoo.com/config/mail?.intl=gr 


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]