Re: [Ayatana-dev] C++ memory allocation

[Ted Gould <ted@xxxxxxxxxx>]

On Tue, 2011-09-20 at 23:27 -0400, Jay Taoko wrote:
> On 9/20/2011 10:48 PM, Ted Gould wrote:
> > On Wed, 2011-09-21 at 11:43 +1200, Tim Penhey wrote:
> >> We can see that calloc is faster only at small sizes.  What's more, the
> >> overhead of using a std::vector decreases rapidly as the size gets
> >> larger.
> > And aren't almost all allocations in most programs smaller in nature.
> > Especially GUI programs and toolkits.  The only place I could think we'd
> > even come close to "large" allocations would be in buffer images, and
> > there shouldn't be too many of those.
> >
> > I think that the old adage still holds: If you want something done
> > right, do it yourself.  But often times you're willing to take the
> > penalty and outsource the work to the standard lib to make life easy in
> > non-critical sections.  So "use the appropriate technology for the code
> > in question." ;-)
>
> The section in question creates a buffer of data to load into and OpenGL 
> texture. So the data can be large.
> However, in that location, we are doing 1 allocation and then we exit. 
> Is one really better than the other in this case?

I would say no.  In general, I'd say any discussion over the style of
one allocation is bike shedding.  Who ever writes the code gets to
decide :-)

		--Ted

Attachment: signature.asc
Description: This is a digitally signed message part