Path: utzoo!utgpu!news-server.csri.toronto.edu!bonnie.concordia.ca!uunet!zaphod.mps.ohio-state.edu!uakari.primate.wisc.edu!aplcen!jhunix!barrett
From: barrett@jhunix.HCF.JHU.EDU (Dan Barrett)
Newsgroups: comp.sys.amiga.programmer
Subject: Re: How fast/slow is memory DEallocation?
Message-ID: <8522@jhunix.HCF.JHU.EDU>
Date: 1 Jun 91 00:57:30 GMT
References: <8504@jhunix.HCF.JHU.EDU>
Organization: The Johns Hopkins University - HCF
Lines: 34

In article <8504@jhunix.HCF.JHU.EDU> I wrote:
>I have noticed that my [LARGE] data structure is being deallocated extremely
>slowly --- almost excruciatingly so.
>At first, I thought the problem might be Manx's malloc()/free()
>functions, so I switched to AllocMem/FreeMem.  This sped things up a bit,
>but not much.

	The problem is solved.  Guess what it was??  I forgot to 
#include <functions.h>, so AllocMem/FreeMem had no prototypes.  For some
reason, this caused FreeMem to be VERY slow:  probably things were being
implicitly coerced in a bad way....

	Here's a little benchmark to tickle your fancy.  This program
does several thousand allocate/deallocate calls on a data structure with
(appropriately) several thousand nodes.  STACK was set to 100,000 just
in case.

	Using AllocMem/FreeMem:

		run time == 30 seconds

	Using MANX malloc/free:

		run time == 5 minutes, 35 seconds

Well, I was certainly surprised!

                                                        Dan

 //////////////////////////////////////\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\
| Dan Barrett, Department of Computer Science      Johns Hopkins University |
| INTERNET:   barrett@cs.jhu.edu           |                                |
| COMPUSERVE: >internet:barrett@cs.jhu.edu | UUCP:   barrett@jhunix.UUCP    |
 \\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\/////////////////////////////////////