LEDA's Memory management

Thread View: comp.lang.c++.leda

3 messages

[T] Back to Threads [V] Tree View [G] Group Articles

3 total messages Started by P. Mantecon Fri, 18 Aug 1995 09:06

#47

Author: P. Mantecon
Date: Fri, 18 Aug 1995 09:06

14 lines
937 bytes

As we all know, LEDA includes the macro  LEDA_MEMORY(type)  to provide
an efficient memory management  schema.  Basically it allocates memory
in chunks of 2^13 -4 bytes and then slices every chunk in cells of the
appropriate size.

I can  understand  the size of the basic chunk to be a power of 2, but
what  puzzles me is that -4 bytes, so my  question  is the  following:
why the chunk size is precisely 2^13-4 instead of being just 2^13?

Ideas, anyone?

		Patricia
		
P.S. Source code appears in _memory.c & memory.h

Re: LEDA's Memory management

#49

Author: Frank Muzzulini
Date: Fri, 18 Aug 1995 16:36

17 lines
1284 bytes

P. Mantecon (pmanteco@haifa.cern.ch) wrote:
> I can  understand  the size of the basic chunk to be a power of 2, but
> what  puzzles me is that -4 bytes, so my  question  is the  following:
> why the chunk size is precisely 2^13-4 instead of being just 2^13?

malloc() always needs some additional bytes to store some information
about the allocated block. If you 'malloc' exactly 8k, it has to
allocate one block more, than needed to store 8k.
Seems to me, that's why they spare out the 4 bytes.

	Muzz

--
  ___  Frank Muzzulini     ua0y@rz.uni-karlsruhe.de     Muzz@irc
 <*,*>  ... until the colour of a man skin is of no more
 [`-']  significance than the colour of his eyes ...        Bob Marley (War)
 -"-"-

Re: LEDA's Memory management

#50

Author: Heribert Dahms
Date: Fri, 18 Aug 1995 20:25

22 lines
1818 bytes

In <412fhq$j98@nz12.rz.uni-karlsruhe.de> ua0y@rzstud1.rz.uni-karlsruhe.de
 (Frank Muzzulini) writes:

: P. Mantecon (pmanteco@haifa.cern.ch) wrote:
: > I can  understand  the size of the basic chunk to be a power of 2, but
: > what  puzzles me is that -4 bytes, so my  question  is the  following:
: > why the chunk size is precisely 2^13-4 instead of being just 2^13?
: 
: malloc() always needs some additional bytes to store some information
: about the allocated block. If you 'malloc' exactly 8k, it has to
: allocate one block more, than needed to store 8k.
: Seems to me, that's why they spare out the 4 bytes.

If I remember right, this space is for some sort of next pointer from LEDA,
not malloc(), and we had to hack around to increase this from 4 to 8, because
OSF/1 on DEC Alpha uses 64bit pointers.
BTW, <8Kb for data structures is a severe limit nowadays; that's not a PDP11.
(Now I prefer Tcl/Tk, where I can easily have a single string of several
Megabytes... (If someone doesn't now what Tcl/Tk is, see the comp.lang.tcl FAQ))


Bye, Heribert (dahms@ifk20.mach.uni-karlsruhe.de)

Thread Navigation

This is a paginated view of messages in the thread with full content displayed inline.

Messages are displayed in chronological order, with the original post highlighted in green.

Use pagination controls to navigate through all messages in large threads.

Back to All Threads