Subject: Re: [boost] combinations and permutations
From: Dave Abrahams (dave_at_[hidden])
Date: 2011-01-06 04:02:33
At Wed, 5 Jan 2011 15:04:04 -0500,
Howard Hinnant wrote:
> On Jan 5, 2011, at 2:45 PM, Howard Hinnant wrote:
> > It doesn't seem to be any slower! :-)
> Correction: If the compiler can see that the functor always returns
> false it isn't any slower. If it can't (separate compilation unit,
> or whatever), there's about a 12% speed hit.
That is consistent with my understanding of modern optimization
capabilities. I wonder if link-time optimization can quash that 12%
Anyway, for algorithms like I think the separate-compilation case
doesn't matter much.
-- Dave Abrahams BoostPro Computing http://www.boostpro.com
Boost list run by bdawes at acm.org, gregod at cs.rpi.edu, cpdaniel at pacbell.net, john at johnmaddock.co.uk