Mail Archives: djgpp/1993/05/10/13:00:09
On Mon, 10 May 1993 gpk AT physics DOT att DOT com wrote:
> I wonder if a memory allocator that keeps the blocks sorted by size in
> a b-tree might be good. You'd expect to get rapid access to a block
> of the approximately correct size ( ln(N) tests ), and since you could
> use a best-fit (rather than a first-fit) strategy, there would probably
> be less memory fragmentation than the classic list_sorted_by_address
> algorithm. Since blocks could be split, and the remainders put back on
> the b-tree, you wouldn't have the BSD lack-of-recycling problem.
>
> --Greg Kochanski
Sounds good, but a Btree would bring in a lot of overhead, or at least,
more than a simple sorted, linked-list scheme. Depending on the
speed/fragmentation/efficiency differences it may be worth it.
/* Andrew */
"As far as the laws of mathematics refer to reality, they are not certain;
and as far as they are certain, they do not refer to reality"
Albert Einstein
- Raw text -