Path: utzoo!attcan!uunet!husc6!cmcl2!nrl-cmf!ames!elroy!gryphon!sarima
From: sarima@gryphon.CTS.COM (Stan Friesen)
Newsgroups: comp.lang.c
Subject: Re: alloca wars
Message-ID: <5338@gryphon.CTS.COM>
Date: 7 Aug 88 21:50:21 GMT
References: <696@ns.UUCP> <697@goofy.megatest.UUCP>
Reply-To: sarima@gryphon.CTS.COM (Stan Friesen)
Organization: Trailing Edge Technology, Redondo Beach, CA
Lines: 18

In article <697@goofy.megatest.UUCP> djones@megatest.UUCP (Dave Jones) writes:
>From article <696@ns.UUCP>, by ddb@ns.UUCP (David Dyer-Bennet):
>> If you don't have that, what advantage is there over malloc or
>> any of the standard allocation routines?
>
>Speed.  The malloc() on in Sun3 OS, for example, is incredibly slow
>after a few thousand blocks have been allocated.

	Indeed, and not just on Suns either. A similar thing happened to
me on a Convergent. In fact it eventually got sooo slow that our customers
complained! (Luckily the large number of allocations was unnecessary, the
code was not freeing a message block, so it was easy to fix by just adding
a free() after the message was displayed. But still a 10-20 fold increase in
speed *just* by adding one free!!)
-- 
Sarima Cardolandion			sarima@gryphon.CTS.COM
aka Stanley Friesen			rutgers!marque!gryphon!sarima
					Sherman Oaks, CA