Re: [boost] High Cost of MS "Safe" STL for Release Builds

From: Simon Francis (thejello_at_[hidden])
Date: 2008-04-12 18:24:40

 I maintain a large body of legacy code that was written back when vectors
were fast. Modifying this old code scares me but the approximately three
times slowdown that we have seen is unacceptable. This application is
noticeably slower on a brand new computer than it was a few years ago.
Switching this code to pointers is something I really want to avoid but this
seems like an all or nothing sort of thing and we rely on third party
libraries beyond our control. This is a really ugly situation we have been
forced into.

On Fri, Apr 11, 2008 at 6:28 PM, Emil Dotchevski <emil_at_[hidden]>
wrote:

> The overhead of the runtime checks is usually most critical in
> operations on vectors; however, vector elements are sequential in
> memory and in performance-critical parts of the code it is trivial to
> use pointers instead of vector iterators. I would much rather do that
> instead of relying on the compiler to get it right, regardless of
> whether runtime checks are enabled or not.
>
> At the same time, I do like the runtime checks in contexts that are
> not performance-critical.
>
> --
> Emil Dotchevski
> Reverge Studios, Inc.
> http://www.revergestudios.com/reblog/index.php?n=ReCode
> _______________________________________________
> Unsubscribe & other changes:
> http://lists.boost.org/mailman/listinfo.cgi/boost
>