Re: Eu 3.1.1 Windows -- memory bug?
Matt Lewis wrote:
>
> Andy Serpa wrote:
> >
> > This program eats up a bunch of memory. It causes my free RAM on my meter
> > to
> > drop by 667 MB. But once the counter gets to 50 and I hit return (once),
> > which
> > replaces all that with a 1-element sequence; I only get 90 or so MB back in
> > version 3.1.1. In version 2.5, I get it ALL back at this point. In v3.1.1
> > it isn't freed (apparently) until the program exits completely.
>
> I believe that what you're seeing is dependent on windows. The way it
> works is that a program asks the OS for some heap space, and the OS allocates
> it in chunks. Then, you can allocate memory within the heap space. Later,
> you can free that memory. IIRC, depending on certain parameters, the OS
> will grow the heap as your application requires. It may reduce the size of
> the heap when you free the memory from your app. Or it may not. If you
> have significant memory fragmentation, you probably won't get that memory
> back.
>
> I don't think that there has been any significant memory management changes
> for windows since 3.0, but only Rob can comment on 2.5.
>
I'm suspicious... running the same program once with 3.1.1 and then immediately
after with 2.5 should produce the same result (more or less) if nothing had
changed between the two versions (which obviously there had or we would not have
had that other bug). There would be no reason for fragmentation. There has got
to be something more to it...
|
Not Categorized, Please Help
|
|
