> When Linux finishes running a program, it leaves it in memory against the
> possibility that someone will run it again before it needs that memory for
> something else. Same with data files. I'm not clear on which of this is
> "shared", which "buffered", but the memory in those two areas is available
> for use by other programs if needed. That's why "free" reports the second
> line (-/+ buffers/cache) -- because that tells you how much memory is really
> available.
>
> One consequence of this is that any host that has a long uptime will show
> very little "free" memory (by the standard of the first line), because
> caching and buffering cumulates. Unfortunately, these are the numbers
> reported by "top", making that part of its output almost useless.
>
> This caching/buffering business is a real performance plus for Linux, BTW.
> Try running something that involves searching a database (a "locate" will
> serve if you use that database; a "find" on a mounted CD is also a handy
> indicator). DO it once and notice how long it takes. Then do the same thing
> again, immediately, and see how much faster it is.
>
> Ray Olszewski
> Palo Alto, CA [EMAIL PROTECTED]
I've already seen the performance increase because of this
behavior, I just didn't relate it to what 'free' reported ... Thanks! If I
start Wine for the second time, it usually comes up much faster
than the first time. That's the reason I'm wondering if stripping could
make Wine faster. After all, 60MB of code can't be read in just one
second ... (I don't really know what this is, I just read about it in the
docs and know that it reduces a program's size by eliminating
debugging info? ...)
Jerem�as Galletti
