This is the mail archive of the gcc-patches@gcc.gnu.org mailing list for the GCC 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: [PATCH] Speed up hybrid_search_bitmap


Hello,

> >maybe, but you will never make it as fast as simple ordinary heaps (the
> >reason to use Fibonacci heaps would be if you used decrease key 
> >operation,
> >as here really matters that you can do it in constant time).
> 
> If this were true, people wouldn't use fibheaps in priority queues in 
> general,
> since decreasekey isn't used that often.
> i've seen plenty of well performing priority queues using fibheaps that 
> don't even expose decreasekey.
> If what you say were true, you'd not expect to see this.

then maybe all of them are using them wrongly? Neverthless, what I see
is the times I have measured, and the fact that attempting to speed up
Fibonacci heaps would be much more complicated than to write simplish
ordinary ones that are fast enough.

Zdenek


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