Re: Data management algorithm needed

> There are different approaches for the latter. One place to look is the
> linux kernel. IIRC it allocates memory in blocks sized at 2**n. If an
> object goes beyond 2**i bytes, it gets a new block of 2**(i+1) bytes, so
> the emphasis is on rapid scaling instead of space efficiency. I could be
> remembering this wrong, but it's worth looking at for a real-world
> example.

Why is allocating space in blocks of 2**i faster?

>
> > Does anyone know of a good survey of the techniques for solving this
> > class of problems? Perhaps a book or a chapter of a book?
>
> There's a book called
> _Garbage_Collection:_Algorithms_for_Dynamic_Memory_Management. I haven't
> read it, but it gets good reviews. See Amazon.

Thanks. The local Borders has it in stock, so I'll head over and take a look. Received on Mon Jan 07 2002 - 01:23:12 CET